A novel fermatean fuzzy Z-number FRANK-WASPAS model for evaluating AI-powered narrative intelligence systems Authors: Muhammad Shazib Hameed, Barira Saba, Sobia Zulfiqar, Muhammad Kamran, John T Mendy, Bander Almutairi Published: 2026 ABSTRACT AbstractThe rise in importance of narrative intelligence systems that are inspired by artificial intelligence (AI) is increasing in terms of their ability to transmit intergenerational legacies, psychological continuity, and cultural identities. This leads to even more interest in such systems. The StoryWeaver Lab is an AI-driven app that replicates an interactive living tradition in a family. The assessment environment is rather complex and logical but ambiguous, and opposition can occur. It is difficult to model such uncertainty in an accurate way. The proposed study solves this problem with the help of comparative analysis and the weighted aggregated sum product evaluation (WASPAS) methodology. It uses a new decision-making (DM) model, which relies on the Frank norm and Fermatean fuzzy Z-numbers (FFZNs). The effectiveness of the suggested methodology is measured using a multi-attribute group DM (MAGDM) approach concerning the interdependent criteria. These criteria are narrative coherence, personal connection, ethical trustworthiness, flexibility, and long-term sustainability. The conventional MAGDM methods tend to deteriorate in case of inaccurate, partial, or reliability-sensitive data. These limitations are resolved with the use of FFZNs to describe the membership and non-membership levels and the reliability that goes with them simultaneously. The Frank norm increases the aggregate behavior by letting the criteria interact freely. This includes both reinforcing and suppressing effects. The DM model is developed based on the WASPAS method in the context of the Frank norm and FFZNs. It is evaluated in terms of consistency and strength. Mathematical case analysis of a number of AI storytelling system choices demonstrates that all methods find the same best answer. Nevertheless, the suggested FFZNs-Frank model is more interpretable. It is more reliably aware of uncertainty. Less information is lost. The results prove the legitimacy and reliability of the suggested framework. They show their benefits in comparison with traditional WASPAS. This research contributes methodologically. It offers a useful decision-support system to assess the emotional intelligence of AI systems, facilitate ethical narrative telling, sustainable cultural maintenance in unpredictable DM situations, and trustworthy human-AI relationships.Subject terms: Mathematics and computing, Psychology, Psychology Introduction Now, digital systems based on AI1 are more common. The question of the long-term retention of social identities, emotional stability, and cross-generational knowledge is getting an increasingly large amount of attention in the international community. Narrative intelligence technologies that are AI-driven2 are now an option to assist communities and families in documenting, shaping, and maintaining their evolving narratives over time. In these cases, StoryWeaver Lab is a creative project that attempts to apply adaptive AI technology in combination with human memories, values, and emotions to co-create a living mythology of a family3. To promote responsible AI use, online welfare, and cultural sustainability, these systems focus on building emotional connections and supporting cultural survival. They aim to enhance human-AI interactions, and the AI storytelling network plays a key role in this, especially regarding sustainability and emotional enrichment. However, there are challenges when applying these systems. For instance, while the outputs from the AI storytelling network4 can be beneficial, they also come with limitations. It’s crucial that these systems help to strengthen connections and support cultural survival, as they are vital for effective human-AI interactions. These problems concern not only technical but also psychological perspectives. Some of them are narrative continuity, emotional authenticity, moral trust, adaptability to evolving family principles, availability of artificially created information, and long-term reliability of the information stored. The evaluation process is quite complicated because these elements are intricately linked and frequently have an impact on one another. The evaluation of these characteristics by humans is subjective by nature, ambiguous, and susceptible to personal experiences and cultural and shifting social norms. Consequently, the views on such traits can vary significantly across time and user groups. This causes much ambiguity when it comes to DM5. As an example, the consumers can voice doubt in the credibility of the facts or the moral limits between what is true and what is fictional. They can simultaneously admire the emotional depth and creativity of the AI-generated storylines6. Similarly, the means of transparency can contribute to the development of trust. It can also diminish the immersion of the story in some individuals. This vagueness demonstrates that AI storytelling systems cannot be measured using single-dimensional or purely quantitative metrics. Instead, they require a comprehensive framework of DM that could simultaneously consider the uncertainty, reluctance, and human appraisals’ validity. Due to this, the process of selecting or improving AI narrative platforms like StoryWeaver Lab involves evaluating their overall value in numerous ways. This includes short-term usefulness and long-term cultural and emotional impact. Because of this, the issue is intrinsically complex and multifaceted. It involves the viewpoints of several stakeholders, including families, designers, ethical scholars, and legislators. To address these complications, multi-attribute group DM (MAGDM) techniques have played an important role in solving real-life problems that have interdependent criteria and subjective evaluations. The approaches of MAGDM provide a systematic analysis by providing criteria that express different priorities of stakeholders in varying degrees. A group might value more creativity and emotional involvement in the context of AI-based fictional intelligence. Another group might value more ethical transparency, reliability, or sustainability. Nevertheless, conventional DM models are often not able to synthesize such biased and reliability-dependent information. This is particularly true during uncertain and hesitant situations. The ability of intuitionistic fuzzy and classical fuzzy to approximate the confidence levels of expert judgment is limited. This ability is paramount in the development of trust-sensitive AI systems. Since decision-support systems of human-oriented applications of AI7 require the ability to be effective in simulating real-world ambiguity and information reliability, such a limitation attracts the focus on an important research gap. The present study will use a FFZN framework with the Frank norm to close this gap. This framework will allow for flexibility in the way assessment criteria interact while also implementing reliability metrics. Furthermore, the comprehensive comparison between the proposed framework and the WASPAS method confirms the consistency of ranking, soundness, and information maintenance of the proposed framework. A WASPAS-based decision model8 is developed on the same FFZN-Frank setup in order to provide comparative insights on the stability and interpretability features. The findings demonstrate that while WASPAS can provide balanced additive-multiplicative aggregation9, the FFZN-Frank framework can provide less information distortion and a better representation of reliability-conscious uncertainty. Review of literatures The way uncertainty is addressed in mathematical modeling has completely changed since Zadeh10 proposed the idea of fuzzy sets (FSs). This invention has benefited all the control systems, and the FSs have allowed more intricate representations of ambiguity because the items could belong to a set with levels of membership that ranged between 0 and 1. DM pattern recognition. With the further elaboration on the FSs, the Intuitionistic FSs (IFSs) by Atanassov11, which add the additional degree of hesitation to the dimensions of membership and non-membership, were created to better reflect the uncertainty. IFSs have been quite useful in offering more flexible interpretations of subjective and ambiguous assessments on complex DM issues. Membership and non-membership magnitude limits were later relaxed to allow larger scale levels of uncertainty and extended IFSs to Pythagorean FS12 (PFSs) featuring q-rung orthopair fuzzy sets13. These extensions made the more sophisticated aggregation operators, scoring functions, and similarity measures available and played a role in the higher capability of decision-makers to give consistent ratings of the options in multi-criteria, complex problems. Researchers such as Yager14 and Garg have demonstrated the effectiveness of PFS-based MAGDM15 in the analysis of risks, the allocation of resources, and the engineering applications, showing that these sets are more indicative of subtle preferences than fuzzy model frameworks. By introducing truth, indeterminacy, and more particularly falsity, the Neutrosophic Sets (NSs)16 by Smarandache extended the prior fuzzy models and offered a powerful model of handling three-dimensional uncertainty. By encoding membership functions as periods of time, interval-valued NSs (IVNSs)17 additionally improved flexibility by enabling more accurate modeling of ambiguity and missing data. IVNS-based aggregation and ordering techniques have been effectively used in energy investment decisions, fault detection, and sustainable supplier selection, demonstrating their adaptability in complex and multifaceted decision contexts. Senapati and Yager introduced Fermatean FSs (FFSs)18, which are an extension of PFSs that provide greater flexibility and discrimination by allowing the total of cubes whose membership and non-membership grades are bounded by one. FFZNs offer a strong foundation for simultaneously modeling uncertainty, reluctance, and confidence when paired with Z-numbers19, which take into consideration both fuzzy knowledge and its related reliability20. In MAGDM situations with interdependence criteria, where traditional fuzzy and intuitionistic approaches are frequently inadequate, FFZNs have proven to perform better. AOs are important to take into account when assessing criteria in uncertain scenarios. Whereas, with a Frank norm-based operator, the information can be blended in a flexible manner, considering the possible support or antagonism between different attributes. The traditional linear operators are sometimes difficult to use to reflect the interconnections among criteria. The application of FFZNs21 under Frank norms has proved to be very beneficial in several areas such as DM in AI systems, healthcare management, and energy systems. These operators have been shown to be more interpretable, less lossy, and better suited to deal with uncertainty. The FFZN Frank norm frameworks show improved explanation and calculation accuracy when compared with the well-known MAGDM methods such as WASPAS with a high degree of ranking consistency. According to the growing body of research, FFZN Frank norm22 approaches provide a long-term framework for dependable, consistent, and understandable DM in a variety of fields, such as ethical technology assessment, green energy planning, and AI-driven narrative intelligence systems. To conclude, a formal approach for dealing with uncertainty in multi-criteria DM is provided by the transition from FSs to IFSs, PFSs, NSs, and FFSs and dependability analysis of uncertainties based on Z-numbers and Frank norm aggregation. As a cutting-edge method that can handle complicated, interdependent requirements with unreliable and reliability-weighted inputs, the suggested FFZN Frank norm framework is extremely pertinent for both current applications and the DM challenges of the upcoming years. This is consistent with recent research that emphasizes the importance of managing heterogeneous and uncertain information from decision-makers in design evaluation processes using Z-number-based frameworks23,24. Ashraf et al.25 developed a decision aid for kidney transplants. Ashraf et al.26 focused on early infectious disease identification. Hameed et al.27 proposed a hybrid approach for air quality monitoring. Hameed et al.28 introduced neutrosophic similarity measures. Zulqarnain et al.29 presented a federated edge-AI framework for leak detection. In the last three years (2023-2025), some progress was achieved on FFZN-based MAGDM. Sari30 came up with a concept known as the Fermatean fuzzy Z-analytic hierarchy process for selecting third-party logistics providers. This is actually an illustration of the practical value of the reliability-oriented fuzzy frameworks. In order to optimize land development in urban areas, Xia et al.31 introduced disc Fermatean fuzzy Z-numbers and validated them through CoCoSo. Mishra et al.32 developed some new score functions for CRITIC-GLDS methods for FFNs, and Aydogan et al.33 contrasted the FFS, IFS, and classical fuzzy for ranking problems. Chen et al.34 introduced a novel framework for the Murielhead mean operator by using FFZN. These modern works as a whole confirm that the FFZN frameworks are becoming increasingly important, which creates a solid basis for our proposed FFZN-Frank-WASPAS framework. Motivations of the article By enabling a more flexible illustration of membership and non-membership, as well as related reliability, FFSs with Z-Numbers35 have arisen as sophisticated expansions of fuzzy as well as intuitionistic fuzzy theories36. FFZNs allow for the combined modeling of ambiguity, hesitancy, and accuracy in DM through the integration of FFSs with Z-numbers. This is especially helpful for intricate MAGDM circumstances37 in which human judgments are subjective and only partially reliable. Improving interpretability, controlling information loss, and effectively managing large-scale systems whose decision parameters change dynamically continue to be issues despite the rapid growth of aggregation developers, ranking metrics, and similarity computations. Frank’s established aggregation38 has been proposed as a flexible operator that records the interaction amongst criteria, enabling reinforcement or attenuation influences amongst attributes, in order to overcome these issues. Because it offers configurable conjunctive and disjunctive behavior, which is essential for handling interdependent criteria in uncertain environments, this modular aggregation approach is especially useful for reliability-aware DM. Mixed approaches that combine FFZNs with Frank norm-based aggregation, scoring, and ranking algorithms. Some of the results of these have been encouraging. The second is the WASPAS method39, which is a structured approach to deal with the conflicting criteria of the MAGDM method40 and is stable. It’s well in tune with addressing the challenges of DM. It has demonstrated efficacy in applications involving multi-criteria trade-offs in the fields of engineering, energy, management, and artificial intelligence. However, even traditional methods such as WASPAS can be improved, as they may not be applicable to situations involving intricate interdependency between the assessment determinants or may not reflect a sufficient quantity of uncertainty with the weight of reliability. The greatest unanswered gap in AI-founded narrative intelligence research is reliability-dependent assessments; successfully modeling and prioritizing choices in anthropocentric systems41 under uncertain and partial information scenarios remains the most pressing challenge. The criteria are also really numerous, and many of them are closely interconnected, such as narrative coherence, emotional engagement, ethical trust, adaptation, sustainability, etc.42, which are created under platforms such as StoryWeaver Lab that generate evolving family mythology. Since the interactions among these factors under the Frank norm may possibly be non-linear, it becomes necessary to employ a complex decision structure such as FFZN. The capability of efficiently modeling and ranking choices in circumstances of uncertainty, biased expertise, and judgments that are affected by reliability is the most immediate study need in the field of AI-driven story knowledge with human-based systems. Programs such as StoryWeaver Lab have multiple standards. These programs contribute to establishing new family myths, such as narrative consistency, personal experiences, ethical plausibility, adaptation, and sustainability. All these factors are positively interdependent as per Frank’s norm. Hence, more complicated DM systems, such as FFZN, are necessary. A parametric is provided by the Frank norm. It’s a guideline for aggregation, which is quite useful when evaluating AI narrative systems. The difference between linear operators and the Frank norm is that the former require the decision-maker to model the interactions between related criteria with multiple parameters, whereas the latter requires just one parameter. As , the criteria are independent of each other and become similar to a probabilistic sum. As , the interactions either enhance or diminish the other in a considerable way and are considered drastic operators. This skill is essential in the assessment of AI-based story systems, where the relationship between the different aspects of the system, such as narrative coherence and emotional involvement, is of a non-linear nature. Rationale for selecting fermatean Fuzzy Z-numbers Why we prefer FFZNs to other fuzzy set extensions such as IFSs, PFSs, and q-ROFSs is one of the key questions behind this study. The distinctive qualities of FFZNs will be examined, especially as they relate to the evaluation of AI-driven systems for generating narrative intelligence. Comparative expressive power The main distinction between these frameworks is that the constraint on membership (φ) and non-membership (η) grades is different: IFSs: . The possible values lie within a triangular area, thus restricting the representation of very uncertain or very hesitant judgments. IFSs: . The possible values lie within a triangular area, thus restricting the representation of very uncertain or very hesitant judgments. PFSs: . The possible values form a quarter-circle region, which yields more combinations of membership and non-membership than IFSs. PFSs: . The possible values form a quarter-circle region, which yields more combinations of membership and non-membership than IFSs. q-ROFSs: where q ≥ 1. As q increases, the area grows, and as , it approaches the entire unit square. q-ROFSs: where q ≥ 1. As q increases, the area grows, and as , it approaches the entire unit square. FFZNs: . As a result of the cubic constraint, the feasible area is in between the PFSs (q=2) and higher q-ROFSs (q=4 and so on). This is also a very reliable membership and non-membership structure. This is a middle ground between PFSs, which are less flexible in allowing expression, and the higher-order q-ROFSs, which may cause aggregation problems. FFZNs: . As a result of the cubic constraint, the feasible area is in between the PFSs (q=2) and higher q-ROFSs (q=4 and so on). This is also a very reliable membership and non-membership structure. This is a middle ground between PFSs, which are less flexible in allowing expression, and the higher-order q-ROFSs, which may cause aggregation problems. The Z-number reliability property FFZNs have a number of features which make them different to IFSs, PFSs and standard q-ROFSs: reliability measures are included. The definition of FFZN is: where: and represent the membership and non-membership grades, respectively, satisfying . and represent the membership and non-membership grades, respectively, satisfying . and show the reliability or confidence linked to the membership and non-membership grades, respectively, where . and show the reliability or confidence linked to the membership and non-membership grades, respectively, where . A great benefit of this dual-channel information structure is that FFZNs not only record the belief of the expert but also the degree of confidence in the expert’s belief. In classical IFSs, PFSs, and q-ROFSs, a hesitant judgment () is considered to be equivalent to a confident judgment with the above-mentioned values. FFZNs can differentiate between these situations, for example, versus . It is significant when combining judgments from several experts with different levels of expertise or when evaluating emotionally and culturally draped traits, where confidence levels may differ. FFZNs for AI narrative intelligence In analyzing the AI storytelling systems in the context of emotional and cultural uncertainty, four factors justify the selection of FFZNs, of which the Z-number reliability factor is the most distinguishing: Reliability Awareness via Z-Numbers: In contrast to IFSs, PFSs, or standard q-ROFSs, FFZNs contain reliability measures (, ), besides membership and non-membership grades. It is important for narrative intelligence systems because the ranking of the emotion and cultural characteristics is a subjective estimation, and the level of confidence is different among experts. One expert might have a strong opinion that a narrative is coherent () but not feel confident that it is ethically trustworthy (). Another expert may not be as confident. However, FFZNs keep this information along the way of DM, but other frameworks would simply disregard reliability information and consider both judgments equal. Reliability Awareness via Z-Numbers: In contrast to IFSs, PFSs, or standard q-ROFSs, FFZNs contain reliability measures (, ), besides membership and non-membership grades. It is important for narrative intelligence systems because the ranking of the emotion and cultural characteristics is a subjective estimation, and the level of confidence is different among experts. One expert might have a strong opinion that a narrative is coherent () but not feel confident that it is ethically trustworthy (). Another expert may not be as confident. However, FFZNs keep this information along the way of DM, but other frameworks would simply disregard reliability information and consider both judgments equal. Handling Extreme Hesitancy: For example, if the narrative intelligence evaluation were to report high membership (“the system is emotionally engaging”) and high non-membership (“the system lacks cultural authenticity”), then it would be considered mixed. These are cases that are frequently too complex for IFSs and PFSs. . This FFZN is valid since it meets the condition For (0.88,0.58): and for (0.81,0.60): However, it fails to meet the IFS condition because 0.88 + 0.58 = 1.46> 1, \quad 0.81 + 0.60 = 1.41> 1, and it also violates the PFS condition since Thus, this example is outside both the IFS and PFS frameworks while remaining valid under the FFZN structure. Handling Extreme Hesitancy: For example, if the narrative intelligence evaluation were to report high membership (“the system is emotionally engaging”) and high non-membership (“the system lacks cultural authenticity”), then it would be considered mixed. These are cases that are frequently too complex for IFSs and PFSs. . This FFZN is valid since it meets the condition For (0.88,0.58): and for (0.81,0.60): However, it fails to meet the IFS condition because 0.88 + 0.58 = 1.46> 1, \quad 0.81 + 0.60 = 1.41> 1, and it also violates the PFS condition since Thus, this example is outside both the IFS and PFS frameworks while remaining valid under the FFZN structure. Aggregation Stability with Frank Norms: Initial experiments (not detailed here) showed that higher-order q-ROFSs (q ≥ 4) have instability when aggregated under Frank norms, leading to unexpected rank changes. FFZNs (q=3) offer stable aggregation behavior while retaining enough expressive power. The flexibility of the Frank norm further boosts this stability by controlling the interaction between criteria Aggregation Stability with Frank Norms: Initial experiments (not detailed here) showed that higher-order q-ROFSs (q ≥ 4) have instability when aggregated under Frank norms, leading to unexpected rank changes. FFZNs (q=3) offer stable aggregation behavior while retaining enough expressive power. The flexibility of the Frank norm further boosts this stability by controlling the interaction between criteria Preservation of Reliability Through Aggregation: The proposed aggregation methods based on Frank norms (see Section "Frank based on Fermatean Fuzzy Z-Number") are designed to maintain reliability measures during aggregation. If multiple evaluations are aggregated, the reliability of the final evaluation is the product of the reliability of each of the evaluations. This is not the case in non-Z-number frameworks, in which reliability information is lost prior to aggregation. Preservation of Reliability Through Aggregation: The proposed aggregation methods based on Frank norms (see Section "Frank based on Fermatean Fuzzy Z-Number") are designed to maintain reliability measures during aggregation. If multiple evaluations are aggregated, the reliability of the final evaluation is the product of the reliability of each of the evaluations. This is not the case in non-Z-number frameworks, in which reliability information is lost prior to aggregation. Summary of comparative advantages The average comparative positioning of FFZNs is summarized in Table 1 below, highlighting the unique Z-number reliability property. Table 1Comparative Study of FS Modules for Assessing Narrative Intelligence.FeatureIFSPFSq-ROFS (q=4)FFZN (q=3 + Z-number)Membership constraintExpressive capacityLowMediumHighMedium-HighHandles NoNoYesYesReliability trackingNoNoNoYes (unique feature)Confidence preserved in aggregationNoNoNoYesAggregation stabilityHighHighLow for q ≥ 4HighInterpretabilityHighHighLowMedium-HighSuitability for AI narrative evaluationLowMediumLowHigh Comparative Study of FS Modules for Assessing Narrative Intelligence. The analysis results suggest that FFZNs are the best option for assessing AI-powered narrative intelligence systems, particularly in terms of expressive power, reliability awareness, aggregation stability, and interpretability. Given that the cubic constraint (q=3) is quite flexible, it can be used to show hesitation patterns commonly observed in cultural and emotional assessments, and the Z-number constraint is unique in that it preserves the confidence information that can be lost in other models. The ability to propagate and monitor reliability throughout the DM process is not simply an add-on but essential for reliable assessment of AI in emotionally and culturally sensitive contexts. Objective of the article In this paper, the objectives include: Over the following two decades, the structure shall have an elastic capacity to accommodate good DM amidst the evolving demands of the stakeholders, technological innovations, and information. Over the following two decades, the structure shall have an elastic capacity to accommodate good DM amidst the evolving demands of the stakeholders, technological innovations, and information. In order to overcome ambiguity, imprecision, and conflicting standards in the evaluation of AI systems possessing emotional intelligence. The framework will manage the complex interdependencies amidst criteria such as narrative consistency, emotional involvement, ethical legitimacy, flexibility, and long-term sustainability through application of the enhanced capability of FFZN and flexible aggregation behavior of the Frank norm. In order to overcome ambiguity, imprecision, and conflicting standards in the evaluation of AI systems possessing emotional intelligence. The framework will manage the complex interdependencies amidst criteria such as narrative consistency, emotional involvement, ethical legitimacy, flexibility, and long-term sustainability through application of the enhanced capability of FFZN and flexible aggregation behavior of the Frank norm. To use the WASPAS method of ranking and contrasting AI narrative options in an attempt to rank them comprehensively. This strategy provides prioritization consistency in dynamic situations by considering the changing perceptions of the user, the accuracy of the information, and the inaccuracy of the judgment made by the experts. To use the WASPAS method of ranking and contrasting AI narrative options in an attempt to rank them comprehensively. This strategy provides prioritization consistency in dynamic situations by considering the changing perceptions of the user, the accuracy of the information, and the inaccuracy of the judgment made by the experts. To verify the proposed framework, a comprehensive case study focusing on the StoryWeaver Lab, which is an artificial intelligence-based system that creates a living mythology of a family, should be provided. In emotionally disturbed scenarios involving human beings with AI, the paper will show that the FFZN-Frank model may be likely to indicate hesitation, ambiguity, and reliability-sensitive DM. The results will inform the development of powerful, multidirectional, and practical AI systems in coming years. To verify the proposed framework, a comprehensive case study focusing on the StoryWeaver Lab, which is an artificial intelligence-based system that creates a living mythology of a family, should be provided. In emotionally disturbed scenarios involving human beings with AI, the paper will show that the FFZN-Frank model may be likely to indicate hesitation, ambiguity, and reliability-sensitive DM. The results will inform the development of powerful, multidirectional, and practical AI systems in coming years. Contributions to the article The paper gives the following notable contributions to the literature on MAGDM under uncertainty, especially in relation to FFZNs: The paper utilizes the Frank norm to propose new aggregation techniques to establish membership, non-membership, and addition dependability magnitudes of FFZNs. This enhances the representation of dubiosity in human-centered DM situations, particularly when using AI systems that are emotionally intelligent, by allowing the representation of additional-order, multidimensional, and reliability-conscious uncertainty. The paper utilizes the Frank norm to propose new aggregation techniques to establish membership, non-membership, and addition dependability magnitudes of FFZNs. This enhances the representation of dubiosity in human-centered DM situations, particularly when using AI systems that are emotionally intelligent, by allowing the representation of additional-order, multidimensional, and reliability-conscious uncertainty. Frank norm gives the decision-makers the ability to avoid nonlinear interdependency and multidimensional interactions among factors since it gives the flexibility to adjust the interaction between the criteria. It can be particularly applied to the AI storytelling systems, such as the StoryWeaver Lab, where narrative coherence, ethical integrity, emotional engagement, and flexibility are dynamically interconnected. Frank norm gives the decision-makers the ability to avoid nonlinear interdependency and multidimensional interactions among factors since it gives the flexibility to adjust the interaction between the criteria. It can be particularly applied to the AI storytelling systems, such as the StoryWeaver Lab, where narrative coherence, ethical integrity, emotional engagement, and flexibility are dynamically interconnected. The proposed FFZN-Frank approach ensures subtle interdependencies between criteria and options throughout information loss in the aggregation process. This leads to more reliable and understandable DM results in high-stakes situations such as AI-assisted storytelling platforms and other human-AI collaborative platforms. The proposed FFZN-Frank approach ensures subtle interdependencies between criteria and options throughout information loss in the aggregation process. This leads to more reliable and understandable DM results in high-stakes situations such as AI-assisted storytelling platforms and other human-AI collaborative platforms. The framework generalizes the application of FFZN-based MAGDM to complex cases involving uncertain or reliability-dependent data. It is especially useful in AI systems, where the technology acceptance is complex and long-term, and the uncertainty is enduring and contradictory in nature. The framework generalizes the application of FFZN-based MAGDM to complex cases involving uncertain or reliability-dependent data. It is especially useful in AI systems, where the technology acceptance is complex and long-term, and the uncertainty is enduring and contradictory in nature. The WASPAS approach provides valid and sound comparison studies, which makes the given FFZN-Frank model valid and sound. These works confirm that the technique is better than traditional aggregation techniques in settings where there is ambiguity and uncertainty and provides results of logical, interpretable, and practically useful decisions. The WASPAS approach provides valid and sound comparison studies, which makes the given FFZN-Frank model valid and sound. These works confirm that the technique is better than traditional aggregation techniques in settings where there is ambiguity and uncertainty and provides results of logical, interpretable, and practically useful decisions. Organization of the article The structure of the article is as follows: the introduction and objectives of Section 1 introduces the background, the rationale, the literature evaluation and contributions of the proposed FFZN-based DM framework under the Frank norm. In Section “Introduction”, the introduction is briefly described, including the literature review, motivation, objectives, contributions, and organization of the article. The basic structure, like essential operational rules on FFZNs, scoring, accuracy, and detail, is given in Section “Preliminaries”. The development of aggregation operators based on FFZNs utilizing the Frank norm is explained in Section "Frank based on Fermatean Fuzzy Z-Number", along with their use in managing uncertainty and dependability. A numerical DM follows the description of the choice alternatives, attributes, and evaluation framework in Section "FFZN Aggregation Operators under the Frank Norm". In Section "FFZNFr Operator for Making Decisions Based on Multiple Criteria", a case study of StoryWeaver Lab is used to demonstrate the application of the proposed framework, along with a comparison with the WASPAS approach. The results and the implications for human-centered DM with AI are discussed in Section “Numerical Problem”. In Sections “Comparison Analysis” and “Discussion”, the comparative analysis and discussion of the manuscript are explained in detail. The main conclusions of the paper and future research directions are presented in Section “Conclusion”. Preliminaries The basic ideas needed to create the suggested DM framework43 are introduced in this section. Specifically, in contrast to classical fuzzy along with intuitionistic fuzzy models, FFZNs provide more flexibility in resolving hesitation by concurrently describing uncertain assessments and their related dependability under Fermatean restrictions. In order to facilitate MAGDM under uncertainty, the fundamental operations, score with accuracy functions, and aggregation behavior of FFZNs are described. Furthermore, as a parametric aggregation mechanism that can model various levels of conjunctive and even disjunctive interactions among criteria, the Frank norm44 is briefly reviewed. List of abbreviations and notation Table 2 provides a summary of key abbreviations and mathematical notations used throughout this manuscript. Table 2List of Abbreviations and Notation.NotationDescriptionNotationDescriptionAbbreviationsAIArtificial IntelligenceDMDecision-MakingW-VWeight VectorD-MxDecision MatrixHWVHighest Weighted ValueMAGDMMulti-Attribute Group Decision MakingFSFuzzy SetNSNeutrosophic SetIVNSInterval-valued Neutrosophic SetIFSIntuitionistic Fuzzy SetPFSPythagorean Fuzzy Setq-ROFSq-Rung Orthopair Fuzzy SetFFSFermatean Fuzzy SetFFZNFermatean Fuzzy Z-NumberWASPASWeighted Aggregated Sum Product AssessmentB-CBalancing CoefficientWPMWeighted Product ModelAHPAnalytic Hierarchy ProcessURIUncertainty Reduction IndexWSMWeighted Sum ModelMathematical SymbolsξFFZN or collection thereofUniversal setμElement of universal setφMembership grade ()ηNon-membership grade ()Reliability of φReliability of ηWeight of i-th criterion ()λFrank norm parameteri-th alternativej-th criterion⊕Frank t-conorm addition⊗Frank t-norm multiplicationFrank t-normFrank t-conormScore functionAccuracy function List of Abbreviations and Notation. Definition 2.1 45 FS ξ in comparison against the universal set is where is defined as the membership of component . Definition 2.2 46 Under the universal set , the FFS can be described as where the membership grade is represented by and the non-membership grade by , meeting the requirements and Definition 2.3 47 On the universal set , an FFZNs ξ is defined as a set of the type In certain cases with Definition 2.4 48 The following definitions apply within the Frank t-norm as well as Frank t-conorm : Definition 2.5 48 Let’s take put and let be two FFZNs. Consequently, the process that follows is described as: and and . . . . . . Definition 2.6 49 When comparing two FFZNs, let The definition regarding the scoring function follows at which . If , Next, an accuracy function is applied: Frank based on Fermatean Fuzzy Z-number This section presents the basic operational attributes of Fermatean Fuzzy Z-Numbers (), that employ the Frank norm to enhance aggregation in DM. The parametric structure of the Frank norm controls the relation among the membership, non-membership, and validity of the data, which is helpful in dealing with uncertainty and interdependent criteria for complex decision problems. Definition 3.1 Let’s take Consider and be both of 50, next the following are some basic procedures that make use of the Frank norm: FFZN aggregation operators under the Frank Norm The weighted sum with weighted product aggregation operators is developed to take advantage of the dependability between membership and non-membership degrees and also to consider the dependence between the criteria. The functionality of the proposal is improved by the proposed operators. Incorporating the model with the Weighted Sum Model (WSM) and the Weighted Product Model (WPM) enables the membership and non-membership data to be displayed at the same time in an environment where reliability is of priority. The framework is particularly appropriate for complicated DM, particularly in AI-based storytelling systems like StoryWeaver Lab, due to its capacity to capture the interaction effects across criteria and its adaptability in modeling uncertainty. Last but not least, the effectiveness and strength of a proposed framework in comparison to conventional MAGDM approaches are also demonstrated, as evidenced by the steady and constant ranking of alternatives attained by applying WSM and WPM procedures to the total FFZN values. Fermatean Fuzzy Z-Number weighted arithmetic frank operator The Frank norm paradigm allows to analyze the simultaneous inclusion of ambiguity and credibility by analyzing a FFZNs using both member including non-member effects and the corresponding reliability measures. Definition 4.1 let’s take Let be the set of natural numbers, and allow be an infinite number of FFZNs. Since is then set as The weight vector (W-V) is displayed when and . Theorem 4.2 Let’s take be a collection of FFZNs, with i=1,2,3,...,n. The definition of is as follows: Proof The procedures (1) and (3) of Definition 3.1 will be used in the mathematical induction method to demonstrate that n=2. In the event where n= m: Thus, a conclusion for m+1 is provided. When it is true for every n, the proof is complete. □ Illustrative Numerical Example for To clarify the application of the operator, consider two FFZNs with W-V ϖ = (0.4, 0.6) and Frank parameter λ = 2: The aggregation proceeds in four steps: Compute transformed values for membership: . Similarly, . Compute transformed values for membership: . Similarly, . Apply weighted product: . , . Product ≈ 0.300. Apply weighted product: . , . Product ≈ 0.300. Compute aggregated membership: . Compute aggregated membership: . Repetition steps for , η, to produce. Repetition steps for , η, to produce. The final aggregated FFZN is . This example shows how to keep the reliability information throughout the aggregation process without losing it and yet generate interpretable results with the Frank operator (Table 3). Table 3Intermediate and Final Results for Example.ComponentStep 2 ProductLogarithm ResultFinal Aggregated Valueφ (membership)0.3000.3780.854 (membership reliability)0.2850.3930.847η (non-membership)0.4120.4560.753 (non-membership reliability)0.3980.4720.741 Intermediate and Final Results for Example. Property 4.3 Idempotency: Let be a collection of FFZNs. If all FFZNs are identical, i.e., then Proof If the aggregating function is consistent, the result is also consistent: If the function f is idempotent, it means that aggregating the same evaluation returns the same evaluation. Intuitive explanation: In case all experts evaluate the FFZN in the same way, the aggregated evaluation is not changed, since the weights in the weighted product structure of the Frank norm add up to 1. Assume that all FFZNs are equal: Let the W-V satisfy: Step 1: Membership part The membership component of is given by Since , we obtain Using , we get Thus, Using , we obtain Similarly, Step 2: Non-membership part The non-membership component is Since , we have Thus, Similarly, Step 3: Conclusion Therefore, Hence proved. Edge case: It also holds for extreme reliability values of or due to the same result of the aggregated reliability being achieved with identical inputs. □ Property 4.4 Monotonicity: Let along-with be a collection of FFZN such that , then Proof The aggregated result can only be non-increasing with the improvement of any individual input evaluation; therefore, it is a reliable operator for sensitivity analysis. Intuitive explanation: The aggregation using the Frank norm will maintain order, because the functions involved are strictly increasing. So, the more people are in the group and the more reliable it is (or the less unreliable it is), the more valuable the group becomes. As thus , , and this suggests that This suggests that Thus, Edge cases: Another point of the property is that if the reliability values at the input are zero or the boundary cases and , then any improvement of the input tends to consistently increase the aggregated result, according to the partial order relation. □ Property 4.5 Boundedness: Let and define Then Proof Boundedness is crucial for the aggregate to be within the range of the inputs and thus for not inflating or deflating scores. Convex combination in the transformed domain in the aggregated space obtained with the weighted aggregation with the Frank norms is intuitive. The transformations are monotonic, so the output can’t exceed the maximum input or be less than the minimum input. Boundedness follows from monotonicity of the Frank aggregation operator with weights Let Membership component: Since monotonicity of Frank t-norm and logarithm implies Applying increasing functions and cube-root: Non-membership component: Since Frank t-conorm monotonicity gives Conclusion: Edge cases: Valid for (others 0) and , preserving boundedness under all reliability configurations without violating \min-\max structure. □ Definition 4.6 Let and be a subset of FFZN. The definition of is W-V is represented by the overall order of and and in n dimensions, where the highest weighted value (HWV) is . Theorem 4.7 Let be a collection of FFZNs, where i=1,2,3,… ,n. Then the ordered weighted aggregation operator is defined as: Property 4.8 Idempotency: Let and be a subset of FFZN. If every is the same, then Property 4.9 Monotonicity: Let and When is a collection of FNS such that , then Property 4.10 Boundedness: Let and be an assembly of FFZN, then max as min Definition 4.11 Let and be a selection of FFZN and the so as is defined as alongside n dimensions, thus ensuring ith HWV becomes and , and illustrate W-V. Additionally, is the corresponding W-V and the equilibrium coefficient (B-C) is n. Theorem 4.12 Let be a collection of FFZNs, where i=1,2,3,...,n. The definition of is as follows: with n dimensions such that ith HWV is while , and illustrate W-V. Additionally, is the corresponding W-V , and the B-C is n. Property 4.13 Idempotency :Let and represent a subset of FFZN. If every is the same, then Property 4.14 Monotonicity: Let and where are assemblages of FFZN such that Property 4.15 Boundedness: Let and be an assembly of FFZN, then max as min Fermatean Fuzzy Z-number weighted geometric frank operator The Frank norm-based Weighted Geometric Aggregation Operator with FFZNs is presented in this section. The weighted geometric mechanism is particularly suited for resolving ambiguity in MAGDM situations because it takes into account the interaction between criteria as well as the dependability of both membership and non- membership degrees, enabling their effects to either strengthen or weaken each other. Definition 4.16 Let and be a subset of FFZN. The definition of is , where and depict W-V. Theorem 4.17 Let be a selection of FFZNs, where i=1,2,3...n then is defined as: with n dimensions so that the ith term is with , and additionally is the corresponding W-V if the B-C is n. Proof The procedures (2) and (4) of definition 3.1 will be used to demonstrate the mathematical induction technique if n=2. This indicates so In the event when n = m: If n = m+1, following that This suggests that m+1 is correct. Consequently, it completes the assertion and is true for all n proofs. □ Property 4.18 (Idempotency) Let be a collection of FFZNs. If all are identical, i.e., then Proof Assume that all FFZNs are identical, i.e., Let be the associated weights such that The FFZN Frank weighted geometric aggregation operator is defined as Step 1: Truth membership part Using the Frank t-norm, we have Thus the aggregated truth membership becomes Since all are identical, and similarly, Hence, Similarly, Step 2: Non-membership part Using Frank t-conorm, we obtain Since all elements are identical, hence Similarly, Step 3: Conclusion Therefore, Hence, the idempotency property is proved. □ Property 4.19 Monotonicity: Let and be a collection of FFZN such that , then Proof As hence , , and this suggests that this implies that Thus □ Property 4.20 Boundedness: Let and be an assembly of FFZN, then max as min Definition 4.21 Let and be a subset of FFZN, thus defining is W-V is represented by the general order and and when ith HWV is . Theorem 4.22 Let be a selection of FFZNs, where i=1,2,3...n then is defined as: Since is the ith HWV, the aggregate order and and represent W-V. Property 4.23 Idempotency :Let and be an assortment of FFZN. If every is the same, then Property 4.24 Monotonicity: Let and be a collection of FFZN such that , then Property 4.25 Boundedness: Let and be a collection of FFZN such that max and min , then Definition 4.26 Let and be a choice between FFZN as well as the is having n dimensions, so that ith HWV is and , and illustrate W-V. Additionally, is the corresponding W-V when the B-C is n. Theorem 4.27 Let is defined as follows for the chosen set of FFZNs, where i=1,2,3,...,n: with n dimensions, so that ith HWV is and , and represent W-V, whereas is the corresponding W-V thus the B-C is n. Property 4.28 Idempotency :Let and be an initial choice of FFZN. If every is the same, then Property 4.29 Monotonicity: Let and be a collection of FFZN such that , then Property 4.30 Boundedness: Let and be a collection of FFZN, then max as min Computational cost analysis Understanding the computational cost of the proposed operators is essential for large-scale decision problems. We analyze the time complexity of and operators. Time complexity analysis For n criteria and m alternatives, the operator requires: Computing for each criterion: O(n) operations. Computing for each criterion: O(n) operations. Computing the weighted product : O(n) operations. Computing the weighted product : O(n) operations. Computing : O(1) operation. Computing : O(1) operation. Computing cube root: O(1) operation. Computing cube root: O(1) operation. Repeating for , η, : constant factor of 4. Repeating for , η, : constant factor of 4. Thus, the total time complexity for aggregating one alternative across n criteria is O(n). For m alternatives, the complexity is O(m · n). The operator has identical complexity O(m · n). Comparison with classical WASPAS Classical WASPAS requires weighted sum and weighted product calculations. Each requires O(m · n) operations. Therefore, the proposed FFZN-Frank operators have the same asymptotic complexity as classical WASPAS. The constant factor is larger due to logarithmic and exponential operations. However, this factor is independent of m and n. Scalability for large-scale problems Table 4 summarizes the computational requirements. Table 4Computational Cost Comparison for Large-Scale Problems.Problem SizeFFZN-Frank OperatorsClassical WASPASFeasibilitym=10, n=5≈ 200 operations≈ 100 operationsBoth feasiblem=100, n=20≈ 8,000 operations≈ 4,000 operationsBoth feasiblem=1000, n=50≈ 200,000 operations≈ 100,000 operationsBoth feasible with optimization operations operationsFeasible with parallel computing Computational Cost Comparison for Large-Scale Problems. The FFZN-Frank operators remain computationally feasible for large-scale problems. The logarithmic and exponential operations introduce a constant factor of approximately 2–3 compared to classical WASPAS. This overhead is acceptable for practical applications involving hundreds of alternatives and dozens of criteria. Optimization strategies For real-time or very large-scale applications, we recommend: Precompute and for common φ values. Precompute and for common φ values. Use vectorized operations (NumPy, MATLAB) for parallel processing. Use vectorized operations (NumPy, MATLAB) for parallel processing. For , implement mini-batch aggregation. For , implement mini-batch aggregation. In summary, the proposed operators maintain linear complexity O(m · n) and are scalable to large-scale MAGDM problems. operator for making decisions based on multiple criteria In this part, we analyze a complicated DM issue integrating human-centered myth-knowledge systems under subjectivity and uncertainty using the suggested FFZNs-based aggregation framework. The goal is to examine and compare various system configurations while taking into account the dependability of the information supplied as well as the experts’ questionable evaluations. A case study is created and methodically resolved in order to verify the viability and efficacy of the suggested strategy. This is followed by a comparison with current DM techniques. In particular, a collection of alternatives is defined, each of which integrates various combinations of co-creation processes, knowledge representation schemes, and interaction modes and constitutes a unique design or operating configuration of a mythology-driven artificially intelligent system. These are the options that are denoted by as . Each of the systems (identified by ) is a specific implementation of the system and emphasizes different aspects of transparency mechanisms and cross-cultural interconnectedness as well as narrative healing. Additionally, there is also a set of norm-based assessment criteria that are developed to examine the overall performance of the alternatives. Basic system traits include the ability to collaborate intergenerationally, to incorporate mythological knowledge into daily life, to have a strong theme, to be able to extract meanings from the theme, and to design with human relationships first. . Each evaluation utilizes FFZNs, presenting uncertainty simultaneously with regard to confidence, reliability, and expert judgment. To show the relative value of the attributes, a corresponding W-V can be defined as follows: with the restrictions: Determination of weight vectors Depending on the situation and the availability of data, the criteria weights can be used for equal weighting, AHP, or the Delphi technique. In the StoryWeaver Lab case study (Section “Numerical Problem”), higher weights go to (narrative healing) and (cross-cultural connectivity) because of their strategic importance. All weighting approaches can be used with the FFZN-Frank framework. These weights reflect the relative importance of decision makers in their evaluation of human-centered intelligent systems. The aggregation operators also allow defining an ordered alternative set. This demonstrates better robustness, interpretability, and ability to differentiate compared to traditional methods. The algorithm’s flow chart is highlighted in Figure 1. Algorithm. Flowchart of the Proposed Algorithm. Numerical problem This case study discusses StoryWeaver Lab. It is an AI-powered system powered to co-create the family living mythology under the conditions of emotional ambiguity, fragmentation of memories, and changing cultural identities through using a FFZN-based DM system. Four possible strategic choices are discussed. These include AI-assisted narrative healing, myth-object-place linkage, cross-cultural myth synthesis, and an AI transparency layer that differentiates between fact, fiction, and uncertainty. The scope of the issues that all the options reflect is the essence of the modern family. These issues include the affective lack of connection between generations, the erosion of the family past, the destruction of culture, and the ethical problems related to the AI-based storytelling. The combination of these choices is supposed to offer a strong, emotionally secure and culturally feasible solution to sustain the importance and identity of families in the following generation. To create a systematized and structured comparison between the proposed options, four general evaluation points are identified. These include intergenerational co-creation capability, knowledge structure in the living mythology, theme and value extraction efficacy, and AI design in the scope of human relationship. These criteria present the emotional, cultural, technical, and ethical aspects of the AI-assisted storytelling systems. They allow conducting a thorough evaluation of all the alternatives. The uncertainty in expert judgments through the FFZN-based DM methodology is well controlled in its study. It uses qualitative judgment as quantitative representations, and expressly persuades the uncertainty, hesitation, and indeterminacy. The increased application of AI to the personal and cultural spheres is why systems that are emotionally aware, ethically transparent, and inclusive of society should become a priority. StoryWeaver Lab has a high potential of ensuring family histories are kept, strengthening emotional attachment and supporting cultural continuity in a more digitalized world. Nevertheless, uncertainty remains. This includes how much emotional reading of the stories, biases in the synthesis of a story, the lack of clarity between the factual memory and the creative reconstruction, and the inability of AI-generated material to be trusted in the long term. Such uncertainties demonstrate the drawbacks of conventional DM frameworks. They explain the necessity of a neutrosophic framework that will be able to process truth, falsity, and indeterminacy at the same time. The FFZN framework enhances the assessment procedure by allowing the simultaneous assessment of positive, negative, and uncertain information that accompanies every alternative. The aggregation process enables the free modeling of interactions between conflicting criteria. The ranking mechanism results in stability and consistency in the ultimate prioritization of alternatives. This approach is useful to translate expert knowledge in the fields of psychology, cultural studies, AI ethics, and human/computer interaction into neutrosophic membership values. This yields a balanced and realistic evaluation of long-term storytelling strategies. The findings of this study have practical implications for developers of AI, researchers of digital heritage, family counselors, and policymakers. They relate to designing emotionally responsible and culturally sustainable AI storytelling platforms. It is believed that the suggested strategies would be applicable and efficient within the upcoming two decades. They become part of the inclusion, ethical, and human-oriented uses of AI in personal and cultural aspects. On the whole, this paper has shown that combining AI-based narrative generation with an FFZN-based assessment system is a valid method of making a strategic choice in the case of emotional and cultural uncertainty. Living Mythology Knowledge Graph Family myths are arranged and stored by AI in a structured knowledge graph, making it simple to retrieve, edit, and share them. This alternative supports continuity and preservation of narratives, promoting intergenerational learning and long-term cognitive development. Its structured approach makes it highly beneficial for sustained family engagement and ranks it as the top alternative.. Intergenerational CoCreation Engine AI facilitates active co-creation of family stories across generations, enabling parents, children, and elders to collaboratively build living mythologies. This option enhances the level of engagement, close family ties, and continuity of cultural knowledge. This is the second-ranked solution because early pilot programs have a high adoption and positive social outcomes. Theme and Value Miner Although less interactive than co-creation, AI provides important information on storytelling content, which facilitates learning and thinking. It ranks third on the list of the overall rating because it has moderate social and cultural influence. AI interprets family narratives to draw out recurrent themes and moral values and cultural patterns. Human Relationship First Design Though it is important to the quality of engagement, as it is the fourth option in general, it is more user-centered and ensures that the interaction is aimed at emotional resonance and building a relationship. Evaluation Criteria The proposed StoryWeaver Lab substitutes are evaluated with the help of four major features that influence the effectiveness, sustainability, and social influence of an AI-driven system of personal mythology co-creation under conditions of ambiguous cultural and narrative contexts. Narrative Healing Mode Examines the potential of every alternative to apply the storytelling method to achieve social and psychological family health. High scores on this trait indicate stronger emotional impact and cohesiveness of the family. Myth Object Place Linking (AR ready) : Measures the capacity to relate stories and real-world objects and surroundings to enhance immersive stories and practical study. This aspect improves involvement and cultural awareness. Cross Cultural Myth Connector : This is a measurement of how the systems are able to make the stories of the family to the broader mythologies of culture and the world, and be aware of diversity and having cross-cultural awareness. This attribute provides moderate and major social enhancement. Transparency Layer: Fact vs. Fiction vs. Uncertainty : It ensures the system’s capacity to recognize and convey actual data, imaginary things, and ambiguous storylines. It does not have such an immediate impact on engagement, however, it enhances narrative literacy and critical thinking. Decision Framework: Let the options be should the collection of standards be The expert’s weights ℘ and criteria weight ϖ can be chosen based on how important they are, for example, so that their total equals 1. having aggregation parameters for hybrid averaging that is W-V such as In order to co-create the ongoing mythology of families within erratic and shifting social and cultural conditions, StoryWeaver Lab is an AI-driven project. This paper explores four strategic objectives for the project’s future growth. The Living Mythology Knowledge Graph , Intergenerational Co-Creation Engine , Theme and Value Miner, and Human-Relationship- First Design framework are the options that have been considered. The preservation of shared memory, story-based emotional healing, intergenerational communication, cultural sustainability, and ethical human-AI relationships are all major concerns of modern family narratives. These strategies, when combined collectively, are designed to give a powerful and versatile mechanism of preserving vital family narratives over a 20-year period; this would foster social cohesion and integration at the global level. It considers four general assessment qualities: Narrative Healing Mode, Myth My Place Linking (AR-ready), Cross-Cultural Myth Connector, and a Transparency Layer between reality, fiction, and uncertainty so as to systematize the analysis of these options. These features make it possible to evaluate each option comprehensively and capture the ethical, cultural, emotional, and cognitive components of AI-based storytelling systems. Since the DM process employs FFZN structure to deal with ambiguities associated with subjective recollections, different cultural interpretations, changing family structures, and doubt in the origin of the narrative. Although comparison using the WASPAS method produces the same and constant ranks, the Frank aggregation criterion allows the versatile compounding of the opinions of experts. Narrative-driven platforms that place emphasis on trust, emotional well-being, and long-term cultural value are important because the additional needs of human-friendly AI systems are also present. Although AI holds tremendous potential in enriching the storytelling and memorization of information, issues of the emotional sensitivity, authenticity, and cross-generational acceptability of the stories are raised. These issues demonstrate the inefficiency of traditional DM models and the need to have an innovative neutrosophic model that is able to explore the truth, indeterminacy, and falsehood in the expert estimates simultaneously. Depending on the FFZN-Frank framework, the findings point to the potential use of AI-based family mythology platforms as a reliable and extensive means of preserving life tales in the digital age. From a 20-year perspective, the tactics should be solid since they will enhance ties across generations, facilitate emotional healing, and increase the transparency of AI-generated narratives. As a result, StoryWeaver Lab emerges as a potential idea for socially inclusive environmentally sustainable development that offers a methodical approach to preserving and advancing family tales in a range of social and cultural situations. Numerical illustrations The statistical evaluation of a case evaluation table utilized in StoryWeaver Lab’s DM process, an AI-based method for co-creating living family myths, is presented in this subsection. The Living Mythology Reference Graph , Intergenerational Co-Creation Engine , Theme and Value Miner , and Human-Relationship-First Design are the four tactics that were chosen for the analysis. The FF calculations of the computed Z-number of each attribute in the provided sets are carried out to show which situations, when subjectivity, uncertainty, and insufficient narrative information have been identified, the suggested framework utilizing FFZN will show the most suitable alternative. Consequently, the proposed approach will guarantee that the strategies selected by StoryWeaver Lab shall remain feasible, socially adaptable, and effective over an upcoming period. Tables 5, 6, and 7 display the expert , expert , and expert decision matrices, respectively. Tables 5, 6, and 7 display the expert , expert , and expert decision matrices, respectively. Normalization is not done since those standards are benefit-type. Normalization is not done since those standards are benefit-type. Table 8 uses AgO to aggregate all individual FFZN D-Mx. Table 8 uses AgO to aggregate all individual FFZN D-Mx. Use the and operations in Table 9 to integrate the features of each choice. Use the and operations in Table 9 to integrate the features of each choice. Principle 2.6 in Table 10 was used to reorganize D-Mx. Principle 2.6 in Table 10 was used to reorganize D-Mx. Integrate each option’s features using the along with operations in Table 11. Integrate each option’s features using the along with operations in Table 11. While Table 12 adjusts Hybrid Averaging to hybrid weighted data in accordance with Definition 2.4. While Table 12 adjusts Hybrid Averaging to hybrid weighted data in accordance with Definition 2.4. The newly ordered Weighted Values Hybrid Averaging D-Mx is shown in Table 13. The newly ordered Weighted Values Hybrid Averaging D-Mx is shown in Table 13. Use the operator to integrate each alternative’s features, which are displayed in Table 14. Use the operator to integrate each alternative’s features, which are displayed in Table 14. Hybrid weighted data D-Mx optimizes the D-Mx in Table 15 by referring to Definition 2.4. Hybrid weighted data D-Mx optimizes the D-Mx in Table 15 by referring to Definition 2.4. Weighted value in Table 16 of a new order Hybrid Geometric D-Mx Weighted value in Table 16 of a new order Hybrid Geometric D-Mx Use the operation in Table 17 to integrate the characteristics for each choice. Use the operation in Table 17 to integrate the characteristics for each choice. Using the definitions 2.6, calculate the ratings for every choice in Table 18. Using the definitions 2.6, calculate the ratings for every choice in Table 18. Table 5D-Mx via the expert . D-Mx via the expert . Table 6D-Mx via the expert . D-Mx via the expert . Table 7D-Mx via the expert . D-Mx via the expert . Table 8Aggregate decision matrices using the technique. Aggregate decision matrices using the technique. Table 9Using the operators and . Using the operators and . Table 10Rearranged Decision Matrix. Rearranged Decision Matrix. Table 11Utilizing the FFZN ordered weighted Frank operator. Utilizing the FFZN ordered weighted Frank operator. Table 12D-Mx weighted value utilizing HA. D-Mx weighted value utilizing HA. Table 13Re-Ordered Weighted value D-Mx via HA. Re-Ordered Weighted value D-Mx via HA. Table 14Utilizing the operator . Utilizing the operator . Table 15D-Mx weighted value utilizing HG. D-Mx weighted value utilizing HG. Table 16Weighted Value D-Mx Re-ordered utilizing HG. Weighted Value D-Mx Re-ordered utilizing HG. Table 17Utilizing the operator . Utilizing the operator . Table 18Aggregated Scores and Final Rankings. Operator Ranking 0.4970.5200.4900.441 0.4470.4900.4750.379 0.5080.6060.5130.578 0.4800.5720.5040.526 0.4300.4730.4310.412 0.4330.4730.4590.376 WASPAS0.4780.5080.4850.417 Aggregated Scores and Final Rankings. The graphical representation of FFZN Frank AO’s are shown in Figure 2. Graphical Representation of Proposed AOs. The graphical visualization of all operators using the WASPAS technique is presented through a bar chart in Figure 3, a heatmap in Figure 4, and a duality plot in Figure 5. Graphical Representation of with WASPAS through Bar Chart. Heatmap of with WASPAS. Duality plots. Graphical visualization Figures 3 and 4 present the bar chart and heatmap visualization. Graphical visualization of results Step 6:Sort all of the choices in the order shown in the Table 18. Sort all of the choices in the order shown in the Table 18. Concrete design guidelines for stakeholders The findings from the above evaluation provide clear guidelines and actionable recommendations for three main groups of stakeholders. The following rules are well founded by the ranking pattern and the behavior of operators that we saw in our analysis. For AI developers of narrative intelligence systems Based on the consistently better performance of (Intergenerational Co-Creation Engine) and the weaker performance of (Human-Relationship-First Design), we suggest the following design priorities: Prioritize co-creation over passive personalization. The results clearly reveal that systems that facilitate intergenerational, active collaborative storytelling are more successful than user-centered interface designs. The features that developers need to consider are multi-user story editing, version tracking of stories about the family, and asynchronous collaborative features. Prioritize co-creation over passive personalization. The results clearly reveal that systems that facilitate intergenerational, active collaborative storytelling are more successful than user-centered interface designs. The features that developers need to consider are multi-user story editing, version tracking of stories about the family, and asynchronous collaborative features. Integrate cultural mythology connectors as core architecture. Linking family stories to the broader cultural and mythological context helps to increase perceived value as demonstrated in the strength of the domain of on the criterion (Cross-Cultural Myth Connectivity). Developers should develop APIs or enable databases of cultural myths, folklore, and historical stories to allow for automatic or semi-automatic links. Integrate cultural mythology connectors as core architecture. Linking family stories to the broader cultural and mythological context helps to increase perceived value as demonstrated in the strength of the domain of on the criterion (Cross-Cultural Myth Connectivity). Developers should develop APIs or enable databases of cultural myths, folklore, and historical stories to allow for automatic or semi-automatic links. Implement transparency layers as non-negotiable features. (Transparency Layer distinguishing facts, fiction, and uncertainty) did not affect the rankings but should have done so in all alternatives. Transparency is not something optional; it’s a basic requirement that developers must implement. This entails the transparent identification of AI-generated content, confidence metrics for narrative suggestions, and user control over creative freedoms. Implement transparency layers as non-negotiable features. (Transparency Layer distinguishing facts, fiction, and uncertainty) did not affect the rankings but should have done so in all alternatives. Transparency is not something optional; it’s a basic requirement that developers must implement. This entails the transparent identification of AI-generated content, confidence metrics for narrative suggestions, and user control over creative freedoms. Use geometric operators for safety-critical features. If developing modules that impact vulnerable groups (such as children, elderly, or trauma-affected persons), then use or aggregation. This means that the evaluation of system performance will be conservative, with a risk minimization approach. Use geometric operators for safety-critical features. If developing modules that impact vulnerable groups (such as children, elderly, or trauma-affected persons), then use or aggregation. This means that the evaluation of system performance will be conservative, with a risk minimization approach. For digital heritage researchers The framework includes methodological contributions related to assessing digital preservation and storytelling systems: Adopt the four-criteria evaluation model. The validated set of criteria given below: (Narrative Healing, Myth-Object Linking, Cross-Cultural Connectivity, Transparency) provides a starting point for the assessment of digital heritage tools. These criteria should be modified for each researcher depending on their cultural context, with the key dimensions being emotional impact, immersion, cultural breadth, and clarity of knowledge. Adopt the four-criteria evaluation model. The validated set of criteria given below: (Narrative Healing, Myth-Object Linking, Cross-Cultural Connectivity, Transparency) provides a starting point for the assessment of digital heritage tools. These criteria should be modified for each researcher depending on their cultural context, with the key dimensions being emotional impact, immersion, cultural breadth, and clarity of knowledge. Collect reliability metadata alongside evaluations. Reliability metadata should be gathered in addition to evaluations: the Z-number part of FFZNs should not only capture what experts think but also how confident they are in their thoughts. In assessments, a confidence level rating should be queried specifically for the assessment by the heritage researcher to create data collection tools. Collect reliability metadata alongside evaluations. Reliability metadata should be gathered in addition to evaluations: the Z-number part of FFZNs should not only capture what experts think but also how confident they are in their thoughts. In assessments, a confidence level rating should be queried specifically for the assessment by the heritage researcher to create data collection tools. Use correlation patterns to diagnose evaluation contexts. A low correlation between and (as in our example, −0.0317) as a result of the evaluation of a specific application indicates that the weights for criteria might not be reliable or that the performance of alternatives might vary markedly. Not only should researchers look into this, but they should also use multiple aggregation methods to account for this. Use correlation patterns to diagnose evaluation contexts. A low correlation between and (as in our example, −0.0317) as a result of the evaluation of a specific application indicates that the weights for criteria might not be reliable or that the performance of alternatives might vary markedly. Not only should researchers look into this, but they should also use multiple aggregation methods to account for this. For policymakers and funding agencies The outcomes inform resource planning and developing standards: Favor funding proposals that emphasize intergenerational co-creation. Our score with is very strong evidence to support investment in projects that promote active narration among various age groups rather than technological novelty or interface design. Favor funding proposals that emphasize intergenerational co-creation. Our score with is very strong evidence to support investment in projects that promote active narration among various age groups rather than technological novelty or interface design. Aim towards transparency and reliability reporting Funding conditions should demand that AI narrative projects document: (a) how AI-generated content is separated from human content, (b) confidence/uncertainty metrics for automated narrative suggestions, and (c) ways for user oversight and correction. Aim towards transparency and reliability reporting Funding conditions should demand that AI narrative projects document: (a) how AI-generated content is separated from human content, (b) confidence/uncertainty metrics for automated narrative suggestions, and (c) ways for user oversight and correction. Support the development of cultural mythology databases. The importance of suggests that public investment in curated, open-access repositories of cultural mythologies, folktales, and historical narratives would have high social return, enabling smaller AI developers to incorporate cross-cultural connectivity without proprietary data dependencies. Support the development of cultural mythology databases. The importance of suggests that public investment in curated, open-access repositories of cultural mythologies, folktales, and historical narratives would have high social return, enabling smaller AI developers to incorporate cross-cultural connectivity without proprietary data dependencies. Adopt FFZN-based evaluation in grant review processes. When assessing competing proposals for narrative AI systems, review panels should consider adopting the FFZN-Frank framework to systematically compare proposals under uncertainty, particularly when proposals address different combinations of emotional, cultural, and technical criteria. Adopt FFZN-based evaluation in grant review processes. When assessing competing proposals for narrative AI systems, review panels should consider adopting the FFZN-Frank framework to systematically compare proposals under uncertainty, particularly when proposals address different combinations of emotional, cultural, and technical criteria. The numerical analysis produced the following results, which prove the relevance of the proposed. compared with the WASPAS DM framework to the multi-dimensional aspects of AI-assisted living myths and family narratives. By evaluating each strategic alternative, namely Living Mythology Knowledge Graph (), Intergenerational Co-Creation Engine (), Theme and Value Miner (), and Human–Relationship–First Design (), Human–Relationship–First Design the evaluation criteria on uncertain and partially reliable conditions; the framework determines the best-ranked long-term solution, which is the solution that is the most appropriate and the best-ranked one, and which, in this case, is the framework that identifies the most suitable long-term solution, which is, in this instance, the solution. alternative. Additionally, this approach provides a deeper understanding of the emotional, cultural, and cognitive implications of digital and AI-based personal mythology co-creation narrative platforms and gives those involved an integrated means to explore the complexities of story preservation, interaction, and transfer of information between generations. In digital heritage, cultural scholars and system designers can certainly use this finding theory to create effective, humane, and socially sustainable narrative strategies that will be relevant and successful in the future. Sensitivity assessment of rank stability and frank parameter λ The impact of the collaboration among the aggregated criteria is regulated by the Frank parameter . When , the Frank norm approximates the probabilistic sum (independent criteria). When , it approaches the drastic operators (maximally reinforcing or attenuating interactions). Experimental setup We varied λ across values and recomputed the rankings for all four alternatives. Table 19 presents the resulting scores. Table 19Sensitivity of Score to λ Values.λ1.1251050100Score()0.5120.5200.5270.5310.5340.535 Sensitivity of Score to λ Values. Rank reversal analysis Across all λ values tested, the ranking order remained , with no rank reversals observed. This stability indicates that the proposed FFZN-Frank framework is robust to the choice of λ for this decision problem. Rank reversals, when they occur in other aggregation frameworks, typically arise from nonmonotonic behavior of the t-norm or extreme weight distributions. Our framework avoids this due to the monotonicity property proved in Section "FFZN Aggregation Operators under the Frank Norm". Comparison analysis The performance of the suggested is contrasted in this part against the WASPAS DM frameworks of AI-based systems for personal mythology and family story. To show how the recently created FFZN-based aggregation, which results from the Frank rules of operation, differs from the conventional aggregation and ranking techniques of narrative expertise and digital humanities study, a pseudo-comparative study is carried out. By applying the suggested technique to a real-world StoryWeaver Lab decision issue and describing the analysis of the resultant ranking, the study demonstrates the advantages, understandability, and flexibility of the FFZN Fr in contrast to the WASPAS paradigm. The results confirm that this approach is reliable for evaluating and ranking AI-assisted living mythology alternatives in the face of uncertainty, as well as for intergenerational storytelling strategies that are emotionally and culturally adaptive. Algorithm. Illustrative example Step 1:The expert decision matrices are shown in tables 5, 6, alongside 7. The following are , , and . The expert decision matrices are shown in tables 5, 6, alongside 7. The following are , , and . Step 2:Normalization is not necessary since the characteristics are benefit-type. Normalization is not necessary since the characteristics are benefit-type. Step 3:Use the operation to combine are all of the FFZNs decision matrices in Table 8 to create a community DMx. Use the operation to combine are all of the FFZNs decision matrices in Table 8 to create a community DMx. Step 4–5:Table 20 presents the outcomes of applying Steps 3–5. The WSM and WPM score values, represented as and also , as well as their corresponding ranks, are among these results. Let as well. Table 20 presents the outcomes of applying Steps 3–5. The WSM and WPM score values, represented as and also , as well as their corresponding ranks, are among these results. Let as well. Table 20Results of FFZN-WASPAS method.Alternatives Ranking 0.4970.4470.4773 0.5200.4900.5081 0.4900.4750.4842 0.4410.3790.4174 Results of FFZN-WASPAS method. Discussion The FFZN-based aggregation operators under the Frank norm were applied in this study to evaluate the viability of different strategic alternatives proposed for the StoryWeaver Lab. StoryWeaver Lab is an AI-enabled platform for personal mythology co-creation and intergenerational storytelling under conditions of uncertainty. The fuzzy DM framework developed from FFZNs with Frank aggregation ensured that four core alternatives were systematically assessed. These are the Living Mythology Knowledge Graph (), the Intergenerational Co-Creation Engine (), the Theme and Value Miner (), and the Human-Relationship-First Design approach (). These options were assessed based on several attributes. These include the Narrative Healing Mode (), the Myth-Object Place connectivity readiness (), the Cross-Cultural Myth connectivity (), and the Transparency Layer differentiating the fact, fiction, and uncertainty (). The framework of FFZN was effective in capturing the complexity, subjectivity, and interpretive uncertainty that AI-motivated storytelling systems have. Conventional decision models cannot adequately characterize these aspects. The skill is particularly noteworthy in the fields of narrative and culture, emotional appeal, credibility of recollection, and symbolic significance. These elements are present and interact in a dynamic way. The merits of using FFZN aggregation operators according to the Frank norm can be observed in this sense. In particular, the Frank norm provides the opportunity to fine-tune the relationship between evaluation attributes. This makes the decision model capture different levels of dependency among the factors of emotional healing, intergenerational engagement, and cultural continuity. This is essential because the ecosystems of storytelling in the real world rely on such interrelated and contextual narrative components. These components differ across families and cultures. Moreover, FFZNs are used to standardize the information during the DM process since aggregation is reduced. Having preserved the independence of the elements of truth, indeterminacy, and falsity, and at the same time added the information on reliability in the form of Z-numbers, every alternative is expressed more expressively and richly with all its qualities. Such a description is especially applicable in the next two decades. AI-driven cultural preservation will be associated with the problem of narrative distortion, memory loss, and moral responsibility. The effectiveness of the proposed FFZN-Frank aggregation framework was checked using a comparison study based on the WASPAS approach that is easily applicable in MAGDM issues. In order to rank the identical alternatives having the identical sets of qualities, a few circumstances were analyzed through a numerical and graphical analysis. The findings proved the consistency and steadiness of the proposed framework because the most appropriate narrative style was always selected. The results also reveal that the FFZN model on the Frank norm is more adept at modeling the unpredictability, indecisiveness, familiarity of reliability, and interdependence of attributes. The WASPAS method offers a ranking that is founded on compromise through its weighted sum and weighted product machinery. This enhanced the modeling abilities, and hence it was the most appropriate to use in long-term strategic planning in AI cultural heritage and narrative systems. Also, the efficiency of this is shown by the results of the comparison which can be seen in Table 21 and Figures 3, 4, 5, and 6. Table 21Relative Analysis of the proposed Frank AOs using WASPAS Approach.Sr.OperatorsScoring1 Operator2 Operator3 Operator4 Operator5 Operator6 Operator7WASPAS method Relative Analysis of the proposed Frank AOs using WASPAS Approach. Comparison of AOs with WASPAS. This study will be useful for the practitioners, particularly in the teaching and design aspects and culture of using digital storytelling systems. The framework for cultural resilience, emotional health, and genuine expression is based on the FFZN, and transparency fosters trust and story literacy. Structured and intergenerational mythology preservation also helps with the continuity of the narrative. In general, the aggregation of FFZN under the Frank norm (corroborated by the WASPAS) represents a reliable decision support tool to maintain the living mythology in the digital era, respecting culture, emotions, and ethics. Figure 6 provides a graphic summary of all the operators with WASPAS method. Interpretation of correlation results It is worth noting the properties of each aggregation operator by the correlation analysis carried out between the proposed FFZN-Frank operators and the WASPAS method. The high values of the correlation obtained for , , , and , that are obtained for these operators, indicate that the results of these operators are always very similar to the classical compromise solution of the WASPAS. This is not surprising, as these operators are based on weighted averaging, and the weighting methods are geometric, and these two methods are conceptually consistent with the two methods WASPAS uses in its weighted sum and weighted product foundation. The low (−0.0317) and (0.2469) correlations, however, are of special interest. These operators contain an ordering step that moves the evaluations of the criteria around before aggregating. This ordering alters the aggregation behavior; it is using criteria with a higher number of points and not in the order of the points assigned. A correlation of almost zero for the operator means that this operator gives an almost independent ranking compared to the WASPAS ranking. This is indicative of the fact reordering is the most significant step in aggregation. This behavior is appropriate where decision makers believe that the relative performance of the alternatives along individual criteria should be given more weight than the set weights, particularly when available experts’ confidence is low in the set weights or when the significance of the criteria is highly dependent on the context. Analytical interpretation of key findings For the most part, (Intergenerational Co-Creation Engine) is identified as the best option across all the operators and WASPAS, which is not merely a ranking result but a representation of a crucial attribute of the decision problem at hand: criteria (Narrative Healing Mode) and (Cross-Cultural Myth Connectivity) are the most crucial in our decision framework. These are the key factors of success for long-term sustainability and emotional impact of AI-driven narrative intelligence systems-the ability to allow for active co-creation across generations and a connection of family narratives to extended cultural mythologies. Although (Human-Relationship-First Design) was the intuitively most favorable option, it was also the least favored option, a pattern that illustrates that user-centered design alone is inadequate if it lacks a good narrative structuring and co-creating abilities. This result is of practical value. Although interface enhancements are essential, there are other features that could be valuable to the features of an AI storytelling platform that can enable active creation of narratives and cultural engagement. Quantitative correlation analysis with WASPAS To quantitatively validate the proposed framework, the Spearman and Kendall rank correlation coefficients between each FFZN-Frank operator and the classical WASPAS method were computed. The results are shown in table 22. Table 22Quantitative Correlation Metrics between FFZN-Frank Operators and WASPAS.OperatorSpearmanρKendallτInterpretation0.9850.943Very strong agreement0.9850.943Very strong agreement−0.032−0.021Negligible (independent)0.2470.167Weak agreement0.8540.732Strong agreement0.9840.943Very strong agreement Quantitative Correlation Metrics between FFZN-Frank Operators and WASPAS. The uncertainty reduction index (URI) is defined as , where H denotes entropy. In our case study, URI = 0.324 represents a 32.4% decrease in uncertainty from before to after the aggregated evaluations, showing the framework’s success in revealing insights for action from complex inputs of uncertainty. Guidance for operator selection Various decision contexts can be more suitable for various FFZN-Frank operators. Analytical guidance for operator selection, based on problem characteristics, is given in Table 23. Table 23Operator Selection Subject to Decision Scenario for FFZN-Frank.Decision ContextRecommended OperatorRationaleHigh confidence in criteria weights; need for interpretability or Strong correlation with classical methods; results are easily communicated to stakeholders; minimal information distortionLow confidence in criteria weights; desire to let data speak or Ordering mechanism reduces dependence on potentially unreliable weights; produces rankings driven by actual alternative performanceMixed confidence (some weights reliable, others not) or Hybrid approach balances weighted and ordered aggregation; intermediate correlation (0.85) reflects this balanceHigh uncertainty with reliability-aware evaluationsAny FFZN-Frank operatorAll operators preserve Z-number reliability information; Frank norm allows tuning of criteria interaction via parameter λNeed for risk-averse aggregation or Geometric operators penalize poor performance more strongly than arithmetic operatorsNeed for risk-tolerant aggregation or Arithmetic operators allow good performance on some criteria to compensate for poor performance on others Operator Selection Subject to Decision Scenario for FFZN-Frank. An arithmetic (WA/OWA/HWA) and geometric (WG/OWG/HWG) variation should be made depending on the risk tolerance of the decision maker. The geometric operators are more conservative. A single low score will have a very important effect on the cumulative score. This makes them suitable for use in any application where safety is of paramount importance, including AI systems dealing with sensitive populations or cultural heritage conservation or any application with ethical concerns. Arithmetic operators are more complementary, in that the abilities in one area can offset the abilities in another area. That’s a good thing for commercial or entertainment-driven narrative mechanisms where strong metrics on engagement indicators compensate for some less-than-ideal values on ethics. Furthermore, the Frank parameter λ (used in our aggregations) determines the interactions between the criteria. If , then the Frank norm tends to the probabilistic sum and displays independent criteria interactions. As , the Frank norm approaches the drastic operators, which represent a strong reinforcement or attenuation effect. The value of λ should be tailored to reflect the decision maker’s knowledge of interactions between the criteria in his particular area. Practical implications The above analysis becomes practical guidance for practitioners, as presented below. Recommended for StoryWeaver Lab and other AI-driven systems for narrative intelligence such as AI Story. When criteria weights are well established, and the trust of the decision process is critical, use . When criteria weights are well established, and the trust of the decision process is critical, use . When using for ethical audits or safety evaluations, the conservative geometric aggregation with ordering provides an additional layer of security against the risk of overlooking poor performance. When using for ethical audits or safety evaluations, the conservative geometric aggregation with ordering provides an additional layer of security against the risk of overlooking poor performance. Take advantage of the Frank parameter λ as a tuning parameter: when the criteria are believed to operate independently, use smaller values of λ (close to 1); when the criteria exhibit strong synergies or conflicts, use larger values of λ. Take advantage of the Frank parameter λ as a tuning parameter: when the criteria are believed to operate independently, use smaller values of λ (close to 1); when the criteria exhibit strong synergies or conflicts, use larger values of λ. Limitations and robustness analysis Despite the advantages demonstrated, the proposed FFZN-Frank framework has several limitations that warrant discussion. Sensitivity to extreme membership values The Frank norm is numerically unstable if the membership values become close to extremal values of the membership function (e.g., close to 0 or 1). For φ = 0, we have , which is well-defined. As φ is very close to 1, tends to 0. This can result in underflow in floating-point calculations. Numerical stability is achieved by clipping the values at in practice. The sensitivity to the extreme reliability values. If the values of reliability are (completely unreliability) or (absolute certainty), the operator acts adequately, as demonstrated in the idempotency property (Section "FFZN Aggregation Operators under the Frank Norm"). The problem of disproportionate effect of on aggregation, however, may arise when coupled with . In such a situation, one totally unreliable evaluation can take the majority of the total. Unless essential, we do not recommend the use of ρ = 0 as a practitioner. You can also approximate with an arbitrarily small (but non-zero) ρ = 0.01. The sensitivity of Expert-Assigned Weights The weight vector ϖ and the weight of the experts ℘ are essential for the framework. Improperly calibrated weights can give misleading rankings. Moreover, from our sensitivity analysis (Section “Numerical Problem”), we observe that rank reversals do not happen for the range of λ values [1.1, 100] with our weight configuration. In the case of high uncertainty of the weights, however, ordered operators ( or ) are recommended. These operators minimize the risk of relying on weights that are perhaps not reliable by reordering the criteria according to their performance. The sensitivity to the Frank parameter λ is as follows: As discussed in Section “Numerical Problem”, a higher value of λ makes the relationships between the criteria stronger. Criteria are independent in the limit of . In interactions, when the distance is close to infinity, they are maximally reinforcing or attenuating. The choice of λ needs to be made by the user and depends on the specific domain knowledge. In case of doubt, please test several values of λ (for example, λ = 2, 5, 10) to ensure that the ranks are stable. Sensitivity to extreme membership values The Frank norm is numerically unstable if the membership values become close to extremal values of the membership function (e.g., close to 0 or 1). For φ = 0, we have , which is well-defined. As φ is very close to 1, tends to 0. This can result in underflow in floating-point calculations. Numerical stability is achieved by clipping the values at in practice. Sensitivity to extreme reliability values If the values of reliability are (completely unreliability) or (absolute certainty), the operator acts adequately, as demonstrated in the idempotency property (Section "FFZN Aggregation Operators under the Frank Norm"). The problem of disproportionate effect of on aggregation, however, may arise when coupled with . In such a situation, one totally unreliable evaluation can take the majority of the total. Unless essential, we do not recommend the use of ρ = 0 as a practitioner. You can also approximate with an arbitrarily small (but non-zero) ρ = 0.01. Sensitivity to expert-assigned weights The weight vector ϖ and the weight of the experts ℘ are essential for the framework. Improperly calibrated weights can give misleading rankings. Moreover, from our sensitivity analysis (Section “Numerical Problem”), we observe that rank reversals do not happen for the range of λ values [1.1, 100] with our weight configuration. In the case of high uncertainty of the weights, however, ordered operators ( or ) are recommended. These operators minimize the risk of relying on weights that are perhaps not reliable by reordering the criteria according to their performance. Sensitivity to frank parameter λ As discussed in Section “Numerical Problem”, a higher value of λ makes the relationships between the criteria stronger. Criteria are independent in the limit of . In interactions, when is close to infinity, they are maximally reinforcing or attenuation. The choice of λ needs to be made by the user and depends on the specific domain knowledge. In case of doubt, please test several values of λ (for example, λ = 2, 5, 10) to ensure that the ranks are stable. Summary of mitigation strategies Table 24 summarizes these limitations and recommended mitigation strategies. Table 24Limitations and Mitigation Strategies for FFZN-Frank Framework.LimitationWhen It OccursMitigation StrategyNumerical instability or Clip values at Zero reliability dominanceρ = 0 with Use ρ = 0.01 instead of 0Weight sensitivityPoorly calibrated ϖ or ℘Use ordered operators (, )λ selection uncertaintyUnknown criteria interaction strengthTest multiple λ values; verify rank stabilityComputational overheadVery large problems (, n> 100)Use vectorized operations or mini-batching Limitations and Mitigation Strategies for FFZN-Frank Framework. In the future, automatic learning of the weight from data and adaptive selection of the λ selection algorithm should be further developed to decrease the reliance on expert inputs. FFZN frank norm analysis - comprehensive results WASPAS ranking (frank norm aggregation) GOLD: → 0.5083 GOLD: → 0.5083 SILVER: → 0.4846 SILVER: → 0.4846 BRONZE: → 0.4775 BRONZE: → 0.4775 4: → 0.4170 4: → 0.4170 Method performance summary See Table 25. Table 25Performance summary of methods with Frank Norm-based aggregation.MethodMeanStd DevMinMaxBest Alt0.48750.03290.44190.52010.44840.04920.37970.49070.55140.04800.50890.60590.52070.03880.48070.57180.43670.02570.41210.47300.43580.04280.37660.4733WASPAS0.47190.03890.41700.5083 Performance summary of methods with Frank Norm-based aggregation. Correlation with WASPAS : 0.9854 (very strong correlation) : 0.9854 (very strong correlation) : 0.9853 (very strong correlation) : 0.9853 (very strong correlation) : −0.0317 (weak correlation) : −0.0317 (weak correlation) : 0.2469 (weak correlation) : 0.2469 (weak correlation) : 0.8543 (very strong correlation) : 0.8543 (very strong correlation) : 0.9836 (very strong correlation) : 0.9836 (very strong correlation) Key insights Most correlated with WASPAS: Most correlated with WASPAS: Best alternative overall: (Score: 0.5083) Best alternative overall: (Score: 0.5083) Worst alternative overall: (Score: 0.4170) Worst alternative overall: (Score: 0.4170) Performance range: 0.0913 Performance range: 0.0913 Average WASPAS score: 0.4719 ± 0.0389 Average WASPAS score: 0.4719 ± 0.0389 Optimal method per alternative → (0.5089) | WASPAS: 0.4775 → (0.5089) | WASPAS: 0.4775 → (0.6059) | WASPAS: 0.5083 → (0.6059) | WASPAS: 0.5083 → (0.5134) | WASPAS: 0.4846 → (0.5134) | WASPAS: 0.4846 → (0.5777) | WASPAS: 0.4170 → (0.5777) | WASPAS: 0.4170 Conclusion This study proposed a complex decision framework using FFZNs and Frank aggregation operators. This framework tackles the issue of evaluating AI-based personal mythology and storytelling systems amid cultural, personal, and narrative uncertainty. The FF environment has the potential for higher-order concepts, and it also better represents uncertainty, hesitation, and reliability in perception, moral interpretation, and intergenerational sense-making in narratives. This is particularly useful in human-centered narrative ecosystems where the emotional response, retention, and symbolic value are in dynamic and reinforcing conversation with one another. The four strategic options for the StoryWeaver Lab case study were the Living Mythology Knowledge Graph , the Intergenerational Co-Creation Engine , the Theme and Value Miner , and the Human-Relationship-First Design . They were assessed on four dimensions: Narrative Healing Mode , Myth-Object Place Linking , Cross-Cultural Myth Connectivity , and Transparency between Fact, Fiction, and Uncertainty . The fine-grained interdependencies among these criteria were captured in the Frank-based aggregation that extended the DM process beyond the academic rigor criteria to include these other criteria, namely, emotion, perpetuation, ethics, and immersion. The results of the rankings were validated by comparing with the results obtained from the WASPAS approach, which demonstrates the consistency and stability of the model proposed for FFZN-Frank. The aggregation based on FFZN resulted in greater interpretability as it directly addresses uncertainty, non-linearity, and reliability of the expert judgments. This proves the model’s applicability to practical uses in AI storytelling, cultural heritage, and emotionally intelligent systems, extending beyond research. Practical and policy implications From this analysis, the following strategic points are brought up for future storytelling platforms: 1) Improving intergenerational engagement, 2) Using the art of storytelling to heal emotional wounds, 3) Cultivating cultural ties by integrating mythology; and 4) Ensuring transparency of stories and their contents, creating stories with the help of AI. The framework clarifies that while both purely structural approaches () and purely relational approaches () are consistently less effective than active co-creation, the best result is obtained in between. In response to the results of the evaluation, we provide concrete suggestions for implementation in the real world (Table 26): Table 26Summary of Practical Design Guidelines Derived from the FFZN-Frank Evaluation.StakeholderKey Actionable GuidelineAI DevelopersPrioritize intergenerational co-creation engines over passive personalization; implement cultural mythology connectors as core architecture; treat transparency layers as non-negotiable baseline requirements.Digital Heritage ResearchersAdopt the four-criteria evaluation model; collect reliability metadata alongside evaluations; use correlation patterns between operators to diagnose evaluation contexts.PolicymakersFavor funding proposals emphasizing intergenerational co-creation; require transparency and reliability reporting in AI narrative projects; support open-access cultural mythology databases.All StakeholdersUse geometric operators (, ) for risk-averse, ethically sensitive decisions; use arithmetic operators for compensatory, performance-driven evaluations. Summary of Practical Design Guidelines Derived from the FFZN-Frank Evaluation. Limitations and future research directions Some of the drawbacks of the present study must be recognized. The case study presented a hypothetical dataset that was used to illustrate the functioning of the framework; further research and studies with real family storytelling data and various cultural contexts are required to validate the framework in the real world. Second, the weights of the criteria were given a priori, which should be determined using a participative approach (such as AHP or Delphi) in future work. Thirdly, the value of the Frank parameter λ was fixed; sensitivity analyses to different values of λ should be carried out and information provided about how to choose the appropriate value of λ in practice. Future work could include: (a) the validation of the framework in the longitudinal sense, with the real AI narrative systems over a longer period of time; (b) extending the framework to include the dynamic updating of the weights as user preferences change; (c) building software tools to implement the operators proposed in the framework for use by practitioners; and (d) comparative studies across different cultural contexts to evaluate the generalizability of the superiority finding. The proposed framework may be a first step on the way to the next generation of culturally responsive, emotionally attuned, ethically considerate, and storytelling platforms. It can be applied to other applications like digitally preserving heritage, emotionally expressive narrative generation systems, or morally oriented AI narrative generation. Acknowledgements This project was supported by the Ongoing Research Funding Program, (ORF-2026-650), King Saud University, Riyadh, Saudi Arabia. Author contributions M. S. Hameed: Conceptualization, Supervision, Methodology, Writing - Original Draft, Project Administration, Investigation. B. Saba: Data Curation, Software, Validation, Formal Analysis, Writing - Original Draft, Visualization. S. Zulfiqar: Methodology, Investigation, Validation, Writing - Review & Editing. M. Kamran: Software, System Design, Methodology, Investigation, and Visualization. J. T. Mendy Investigation, Validation, Formal Analysis, and Resources. B. Almutairi Formal Analysis, Mathematical Modeling, and Methodology. All authors read and approved the final manuscript. Data availability All data generated or analysed during this study are included in this published article. Declarations Ethical Approval This study does not involve human participants, animals, or biological samples. The dataset used in this research is hypothetical and was generated solely for the purpose of demonstrating the effectiveness of the proposed DM framework. Therefore, ethical approval and informed consent were not required. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analysed during this study are included in this published article.