PROTEX

Research architecture for structured behavioural case analysis

PROTEX is a methodological research project exploring how artificial intelligence can support the structured analysis of behavioural case material.

The project focuses on designing controlled retrieval architectures and formal case-description frameworks that allow behavioural cases to be represented, queried, and compared while preserving clear distinctions between factual description, behavioural patterns, and interpretation.

This work forms part of a broader investigation into how AI systems can operate on structured knowledge under controlled epistemic conditions.

The PROTEX architecture is not designed to generate predictions or automated profiles. It operates under explicit epistemic constraints, with emphasis on transparency, auditability, and clear evidentiary boundaries.

Research focus

The central problem addressed by the PROTEX project concerns the representation of behavioural case knowledge within AI-supported analytical systems.

Narrative case materials frequently combine factual description, interpretation, and narrative framing within a single text. When processed through conventional retrieval systems, these layers can easily become blurred, leading to uncontrolled inference or misleading analytical outputs.

PROTEX approaches this challenge by treating knowledge representation itself as a methodological design problem. The system architecture separates factual, behavioural, interpretative, and metadata layers, allowing analytical queries to operate within clearly defined epistemic boundaries.

Architecture and methodology

At the architectural level, PROTEX combines semantic retrieval with deterministic metadata filtering and explicit query routing.

Cases are represented as structured analytical objects composed of case profiles, narrative fragments, and contextual metadata. This design enables controlled comparison of behavioural patterns while maintaining traceability between system outputs and the underlying source material.

The objective of the system is not to automate explanation or produce conclusions on behalf of the analyst, but to support disciplined analytical reflection by organising complex case information into transparent, queryable structures.

Research stages

The current stage focuses on controlled fact extraction from a structured repository of historical criminal cases.

The system is designed to extract information from individual cases while enforcing strict evidentiary boundaries, ensuring that outputs remain grounded exclusively in retrieved source material.

Future stages explore expansion of the case repository, cross-case comparison, pattern detection, and structured hypothesis generation.

Throughout all stages of development, human interpretation remains central. The system is intended as analytical infrastructure rather than an automated decision-making tool.

Ethics and design principles

– historical cases documented in public sources
– no operational investigative use
– no automated profiling or diagnosis
– explicit modelling of uncertainty and missing data
– transparent and auditable system architecture

Contact

Email
karol@protex-profiler.ai