Fact-Centric Formal Model and Rules for Incremental Updating of a Repository for Digital Information Flows in Multi-Agent Systems

Abstract

This paper proposes a fact-centric formal model and an incremental maintenance approach for a fact repository to represent heterogeneous digital information streams in a multi-agent environment. The approach covers fact extraction from streams, maintaining a local repository at the agent level, and controlling the fact life cycle via repository operators (insert, update, evidence linking, merge, and close). To ensure repository consistency and verifiability, invariants are introduced for idempotent insertion, temporal ordering, forward-only state transitions, and preservation of audit/evidence traces. In a prototype experiment on an open public-requests stream, 20,000 records were consolidated into 6,319 final facts, yielding a 68.41% data reduction; in 19,544 cases, a resolution description was linked to facts as evidence. The results indicate that the approach provides a practical foundation for monitoring/audit and subsequent indexing and retrieval stages.