KRAFT – A Knowledge-Graph-Based Resource Allocation Framework
Leon Bein, Niels Martin, Luise Pufahl · 2025
Read KRAFT – A Knowledge-Graph-Based Resource Allocation Framework by Leon Bein, Niels Martin, Luise Pufahl (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-2503-21636.txt.
Introduction
Leon Bein, Niels Martin, and Luise Pufahl built KRAFT in 2025, a frame that assigns staff to tasks with the help of a knowledge graph. Firms must weigh roles, skills, case traits, and rules for each open task. Rule lists written by hand are rigid and costly to keep up, while fully auto systems hide why a choice was made.
A knowledge graph stores people, skills, and links in a web that a computer can read. KRAFT draws on that web to propose assignments that fit the case at hand. Each proposal can be traced back through the web, so a manager can see the grounds. The aim is staffing that adapts fast yet stays open to check.
Catalog record
| CREATOR | Leon Bein, Niels Martin, Luise Pufahl |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-21636
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
KRAFT – A Knowledge-Graph-Based Resource Allocation Framework Leon Bein, Niels Martin, Luise Pufahl Abstract Resource allocation in business process management involves assigning resources to open tasks while considering factors such as individual roles, aptitudes, case-specific characteristics, and regulatory constraints. Current information systems for resource allocation often require extensive manual effort to specify and maintain allocation rules, making them rigid and challenging to adapt. In contrast, fully automated approaches provide limited explainability, making it difficult to understand and justify allocation decisions. Knowledge graphs, which represent real-world entities and their relationships, offer a promising solution by capturing complex dependencies and enabling dynamic, context-aware resource allocation. This paper introduces KRAFT, a novel approach that leverages knowledge graphs and reasoning techniques to support resource allocation decisions. We demonstrate that integrating knowledge graphs into resource allocation software allows for adaptable and transparent decision-making based on an evolving knowledge base. 1 Introduction Organizations execute business processes to deliver products or services to their customers. Effective and efficient process execution relies on various resources, including staff, machines, vehicles, and software that perform the work of process activities [ 14 ] . These resources are diverse and have varying capabilities and functionalities. At the same time, they are valuable, often costly, and scarce. During process execution, each newly enabled task, i.e., activity instance, requires allocating one or more suitable resources, a decision that is inherently complex [ 22 ] . For human resources, e.g., factors such as individual expertise, past experience, and team membership, but also their availability, task-specific characteristics, other competing tasks, etc. can influence the speed, outcome, and quality of task execution [ 2 ] . Additionally, dependencies between preceding and subsequent tasks within the process further complicate resource allocation [ 31 ] . We refer to such factors as knowledge that needs to be considered during resource allocation. While resource allocation can be performed manually, information systems and business process management systems (BPMS) typically provide capabilities to support this complex decision [ 14 , 22 ] . In many cases, process experts then define resource allocation knowledge as extensions to a process model, e.g., by specifying rules and constraints for process activities [ 26 , 7 ] , formulating allocation problems and solution techniques [ 17 ] , or incorporating additional visual elements [ 6 ] . Furthermore, process experts must define the set of available resources along with their attributes and, in some cases, the organizational structure, such as hierarchical relationships [ 33 ] . This approach often requires significant manual effort for specification and raises maintenance challenges , as business processes are regularly adapting to changing environments [ 23 ] , also necessitating changes in resource allocation. Approaches have been developed to extract resource allocation knowledge from past process execution data and use this knowledge for real-time resource allocations [ 22 ] . However, such approaches are limited in the concepts for which knowledge can be extracted [ 24 ] . Further, explainability is necessary for complex process-decision-making systems [ 13 ] to build trust in the system’s decisions and is missing in current resource allocation support [ 22 ] . System-provided reasoning allows human decision-makers to make informed decisions or adapt system-made decisions while saving the time to look up the relevant information [ 27 ] . Methods from the fields of knowledge graph reasoning and ontology-based business process modeling appear promising to bridge this gap. Knowledge graph reasoning aims to derive new knowledge from rich graph representations of the real world [ 8 ] , providing tools that can consider all that knowledge in an explainable way [ 11 , 27 ] . In turn, ontology-based business process modeling is concerned with extending classical process models with instances of ontologies that represent additional background process knowledge [ 10 ] , resulting in rich graph structures of process knowledge. Research Objectives While existing works, e.g., propose to consult taxonomies of resource criteria [ 20 , 2 ] or propose the usage of graph reasoning to recommend next activities [ 4 ] , the usage of knowledge graph reasoning in combination with ontological modeling has not yet been investigated for resource allocation in business processes. To this end, this paper investigates this usage, considering the following objectives: section 1 It should support capturing and encoding a broad set of allocation knowledge section 1 It should support an easy adaptation of the encoded allocation knowledge section 1 It should determine (the most) suitable resources based on that knowledge and comprehensively explain the decisions made To realize these objectives, this paper proposes KRAFT, a Knowledge-graph-based Resource Allocation Framework for Tasks in business processes. It allows populating, updating, and reasoning on a knowledge graph that captures resource allocation knowledge for explainable resource allocation decision support. Further, we present a prototypical implementation and demonstration of it on a loan application process, provided as a real-world data set in [ 28 ] . In Section 2 , we discuss preliminaries and related work on resource allocation and knowledge graphs. Then, in Section 3 , KRAFT is presented in detail, and demonstrated using a proof of concept implementation in Section 4 . Finally, we discuss our findings in Section 5 and conclude in Section 6 . 2 Background & Related Work 2.1 Resource Allocation In the business process management (BPM) field, resource allocation describes the decision of assigning concrete process tasks (i.e., enabled activity instances) to resources during process execution, adhering to run-time information (e.g., the resource availability) and previously defined allocation goals and constraints [ 7 ] . In this work, we mainly focus on allocating tasks to human resources. Resource allocation can have various (optimization) goals , such as balancing workloads, or minimizing cost or cycle time. Further, resource allocation must follow hard constraints (e.g., the separation of concerns principle or specific permission requirements) and soft constraints (e.g., preference for resources with a particular experience), where non-adherence may be associated with a penalty [ 22 ] . To avoid lengthy manual decision-making, resource allocation can be supported by information systems, which either fully automate the decision or provide recommendations for human decision-makers [ 33 , 26 ] . These systems need to be aware of knowledge for performing suitable resource allocations. We define allocation knowledge as all this (potentially implicit) knowledge relevant for making an allocation decision. In literature, Ihde et al. [ 17 ] and Pufahl et al. [ 22 ] describe that knowledge about the resource allocation problem is needed: each activity is associated with a certain resource allocation problem with certain goals (contributing to the overall process goal) and allocation constraints (e.g., for complying to specific compliance rules or allowing a certain process quality). For handling these problems, allocation mechanisms [ 33 ] can be selected and are part of the allocation knowledge. This includes allocation rules as collected by Russel et al. [ 26 ] , such as role-based allocation , i.e., requiring assigned resources to possess specific roles or assigning based on previous work on the case, etc. Ouyang et al. [ 20 ] and Arias et al. [ 2 ] discuss knowledge about individual resources such as their roles, their expertise, workload, social context and the organizational setting as well as