The Role of Knowledge Graphs in Simulated Intelligence Analysis: Preliminary Investigation Into Filters and Framing

This study explores the integration of Knowledge Graphs (KGs) into simulated Intelligence Analysis activity, focusing on how structured data representations influence strategies for reasoning. We investigate how different types of KG-based filters—Bloom search phrases aligned to a KG schema (“Analysts’ Heuristics,” “Analysis Procedures”) and Graph Data Science (GDS) algorithms—interacted with two investigative frames: financial transactions (“Following Money”) and criminal activities (“Following Crime”). Participants applied these filters while exploring a scenario-based KG of a simulated criminal investigation, with responses collected through repeated questionnaires and post-task interviews. Results revealed that alignment between filter type and investigative frame influenced consistency and anchoring in participants’ decisions. Anchoring was evident but varied across filter types, with KG-aligned question-based filters supporting earlier convergence (between participants) than Graph Analysis Algorithms. The framing influenced reasoning strategies: the “Following Money” frame prompted more abductive and deductive reasoning, while the “Following Crime” frame led to more inductive generalisations. The findings underscore the importance of filter interpretability and contextual framing in supporting investigative cognition. This work contributes to the design of human–AI collaborative systems by demonstrating how KG schema, perspectives and filters jointly shape analytical outcomes in complex decision environments.