Search papers, labs, and topics across Lattice.
The paper introduces Entropic Claim Resolution (ECR), a novel RAG approach that selects evidence by maximizing Expected Entropy Reduction (EER) over competing answer hypotheses, rather than relying solely on relevance. ECR dynamically selects atomic evidence claims until epistemic sufficiency is reached, as defined by an entropy threshold. Experiments integrating ECR into a production retrieval pipeline demonstrate its ability to resolve epistemic uncertainty in knowledge-intensive tasks by retrieving more discriminative evidence.
Forget blindly retrieving the most relevant documents – RAG systems can achieve better reasoning by strategically seeking out the evidence that most reduces uncertainty about the answer.
Current Retrieval-Augmented Generation (RAG) systems predominantly rely on relevance-based dense retrieval, sequentially fetching documents to maximize semantic similarity with the query. However, in knowledge-intensive and real-world scenarios characterized by conflicting evidence or fundamental query ambiguity, relevance alone is insufficient for resolving epistemic uncertainty. We introduce Entropic Claim Resolution (ECR), a novel inference-time algorithm that reframes RAG reasoning as entropy minimization over competing semantic answer hypotheses. Unlike action-driven agentic frameworks (e.g., ReAct) or fixed-pipeline RAG architectures, ECR sequentially selects atomic evidence claims by maximizing Expected Entropy Reduction (EER), a decision-theoretic criterion for the value of information. The process dynamically terminates when the system reaches a mathematically defined state of epistemic sufficiency (H<= epsilon, subject to epistemic coherence). We integrate ECR into a production-grade multi-strategy retrieval pipeline (CSGR++) and analyze its theoretical properties. Our framework provides a rigorous foundation for uncertainty-aware evidence selection, shifting the paradigm from retrieving what is most relevant to retrieving what is most discriminative.
