EvoWiki: Incremental State Overwriting and Traceable Question Answering for Cross-Meeting Knowledge Evolution
2026-08-24 • Computation and Language
Computation and Language
AI summaryⓘ
The authors created EvoWiki, a system that helps keep track of changing facts during long collaborations, like meetings where decisions are revised over time. Instead of just piling up all past information or picking facts based on relevance, EvoWiki carefully updates which facts are current and which are outdated, keeping a clear history. Their reading method then uses this organized info to give accurate and traceable answers. They also made a new test called CrossMeet to check how well systems handle evolving facts, and found that EvoWiki performs better and is more reliable when facts change frequently.
incremental question-answeringknowledge lifecyclestructured readingentity version chainsState-Overwrite Protocolmulti-hop reasoningfactual consistencytemporal reasoninglong-term collaborationRAG (retrieval-augmented generation)
Authors
Dongsheng Chen, Tianyu Wang, Wenhui Que
Abstract
In long-term collaboration spanning multiple meetings, factual states such as decisions and risks are continually revised, overturned, and replaced. Existing long-context methods typically stack the entire history, while many RAG and structured-memory methods organize knowledge as static or append-only facts and rely on semantic relevance at read time. Without explicit modeling of knowledge lifecycles, these approaches may retain conflicting old and new states simultaneously or discard history, leading to stale retrieval and answers that are difficult to verify. We present EvoWiki, an incremental question-answering architecture for dynamic long-form text. EvoWiki decouples offline incremental construction (BUILD) from online structured reading (READ). BUILD captures the intra-meeting micro-evolution from proposal to decision and uses entity version chains and a fine-grained State-Overwrite Protocol to explicitly distinguish current valid states from superseded history while preserving meeting-level provenance anchors. READ bypasses relevance-based Top-k retrieval and performs deterministic entity addressing, temporal resolution, and cross-entity multi-hop aggregation over the complete Wiki to produce grounded and traceable answers. We further introduce CrossMeet, a high-fidelity bilingual benchmark designed to simulate long-term state evolution, covering factual consistency, temporal reasoning, and cross-meeting multi-hop reasoning. Across six datasets and two reader models, EvoWiki improves macro-average Judge Accuracy over the strongest baselines by 9.72 and 10.00 percentage points, respectively. Human evaluation shows that EvoWiki is more robust and factually faithful under frequent state flips, validating valid-state-oriented reading as a reliable approach to cross-meeting knowledge evolution.