Modeling Interactive Narrative Systems: A Formal Approach

Jules Clerc ( ORCID ), Domitile Lourdeaux ( ORCID ), Mohamed Sallak ( ORCID ), Johann Barbier, Marc Ravaine · 2025

Read Modeling Interactive Narrative Systems: A Formal Approach by Jules Clerc ( ORCID ), Domitile Lourdeaux ( ORCID ), Mohamed Sallak ( ORCID ), Johann Barbier, Marc Ravaine (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-05653.txt.

Introduction

Jules Clerc and colleagues gave interactive stories a formal modeling language in 2025. In interactive narratives, users shape the plot instead of just watching, but research groups each describe their systems differently, which blocks comparison and shared evaluation. Borrowing from across the field, the authors propose one shared vocabulary and structure for describing such systems.

A test case retells Little Red Riding Hood inside the framework to show it working. The formal description made the system's properties clearer and improved how it could be evaluated. A common language should help teams compare designs, reuse parts, and judge what makes an interactive story good. The field gains a step toward shared standards.

Catalog record

CREATORJules Clerc ( ORCID ), Domitile Lourdeaux ( ORCID ), Mohamed Sallak ( ORCID ), Johann Barbier, Marc Ravaine
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-05653

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Modeling Interactive Narrative Systems: A Formal Approach Jules Clerc ( ORCID ), Domitile Lourdeaux ( ORCID ), Mohamed Sallak ( ORCID ), Johann Barbier, Marc Ravaine Abstract Interactive Narrative Systems (INS) have revolutionized digital experiences by empowering users to actively shape their stories, diverging from traditional passive storytelling. However, the field faces challenges due to fragmented research efforts and diverse system representations. This paper introduces a formal representation framework for INS, inspired by diverse approaches from the state of the art. By providing a consistent vocabulary and modeling structure, the framework facilitates the analysis, the description and comparison of INS properties. Experimental validations on the "Little Red Riding Hood" scenario highlight the usefulness of the proposed formalism and its impact on improving the evaluation of INS. This work aims to foster collaboration and coherence within the INS research community by proposing a methodology for formally representing these systems. † † footnotetext: Contact: [email protected], [email protected], [email protected], [email protected], [email protected] Keywords Interactive Narrative System, State transition system, Probabilistic graphical model, Simulations, Formalism 1 Introduction Unlike traditional narratives where viewers are spectators, INS allow users to directly influence the course of a story. These systems find applications in various sectors, including entertainment, education and training, with prominent examples such as “Choose Your Own Adventure” books and interactive systems like Façade [ 1 ] . The field has recently seen new research activity since the arrival of LLM, that excel in a wide range of tasks, have enabled many advances in the field of artificial intelligence, particularly concerning text generation and natural language understanding [ 2 ] . However, despite the arrival of these new revolutionary tools, the field of interactive narrations still suffers from certain problems such as the absence of a formalism commonly used by the entire community. This problem is even more important today, because of the hype caused by these AI, which is leading to the creation of many new INS that are difficult to analyse and compare due to a lack of scientific framework [ 3 ] . To address these challenges, this work proposes a methodology which aims to model and study the properties of INS. By developing systems around a coherent structure, we aim to eliminate ambiguities and provide a basis for future work, such as the possibility of defining and comparing INS properties from a functional analysis. This approach is part of a broader effort by the research community to coordinate and pool knowledge, as highlighted by the Association for Research in Digital Interactive Narratives (ARDIN) [ 4 ] . In the following sections, we detail the proposed formalization, illustrate its application through experiments, and explore its implications for the evaluation and development of INS. 2 Related Work The lack of consensus extends to terminology, which varies across application contexts such as gaming, training, or education. To ensure clarity and consistency, we propose a set of definitions. • Player : The individual who engages with the system and performs actions. In some literature, this role is referred to as the "user" or the "trainee." • Creator : The individual responsible for designing the system, including its states, transitions, and overall structure. Other terms for this role include "developer" or "author." • Experience Manager (EM) : A system-side "player" who has complete control and knowledge of the narrative, enabling the orchestration of events and experiences. This role is sometimes described as the "coach" or "trainer" in contexts involving training scenarios. Despite the progress made to make interactive narratives better, the lack of consensus is still a notable issue in this this field of research. Early work in interactive storytelling focused on the comparative analysis of INS [ 5 ] and notes that these systems are often evaluated individually and are not rigorously compared to other approaches. This can be explained by the absence of an established benchmark (No Yardstick) [ 6 ] . One of the difficulties is that they use different stories. Since then comparisons work has been done, as between the ASD and PAST systems [ 7 ] , but the fact that PAST was implemented on top of ASD, and uses the same story, makes comparisons easier because it’s based on the same representational basis. Despite the fact that the diversity of representation methods influences the difficulty of comparing these systems [ 5 ] , there are studies comparing experience managers with heterogeneous designs like the one based on the "Generalized Experience Management" (GEM) framework [ 8 , 9 ] . This structural analysis is based on the claim that all EM can be represented using a subset of the 5 blocks described by the framework and doesn’t rely on the fact that the systems analyzed were explicitly designed with GEM in mind. After that, each block is analyzable and comparable after decomposition. Other comparisons are made by adopting a functional (or qualitative) approach rather than a structural one by examining properties [ 10 , 11 , 12 ] . Our aim is to improve the INS comparison using a functional approach, based on a formalism capable of defining systems and their associated properties. For a better understanding, structural analysis can be seen as a "white box", which focuses on the internal structure of the system in order to know the modules that make up the system and how they are developed, as does the GEM framework [ 9 ] . Conversely, functional analysis can be seen as a "black box", because it doesn’t have access to the structure of the system. Its goal is to analyze the system according to the specifications it must respect, defined by properties. In summary, structural analysis looks at how the system is built, while functional analysis looks at how the system behaves. All these approaches have the advantage of being modular and to combine related research. This allows research to progress by pooling the work of different researchers, as it was the case for the PAST system with the structural analysis from the GEM [ 7 , 9 ] . Still with the aim of comparison, GEM researchers are working on developing a complete evaluation plateform for EM (currently under development on github) [ 13 ] . But even if a platform is finally released, it’s still necessary for researchers in the field to adopt this tool and the associated representation which can be time-consuming, because the representation of the system must often be done in parallel with the design [ 14 ] . Furthermore, the systems that the researcher wishes to compare must also be converted and few of them are released after the publication of the linked scientific article [ 13 ] . However, we assume that this effort is necessary, in order to coordinate and pool knowledge, enabling future research to be improved. Our work is part of the effort to seek a common representation that can be used for all types of system design, or also for researchers wanting to coordinate their work and compare their results around a common framework. More specifically, we focus our efforts around a functional analysis methodology, by defining and comparing the system properties. Current systems are designed for a specific domain [ 9 ] in relation to their

Read the complete text