DeepRefine: Agent-Compiled Knowledge Refinement via Reinforcement Learning
Agent-compiled knowledge bases provide persistent external knowledge for large language model (LLM) agents in open-ended, knowledge-intensive downstream tasks. Yet their quality is systematically limited by \emph{incompleteness}, \emph{incorrectness}, and \emph{redundancy}, manifested as missing evidence or cross-document links, low-confidence or imprecise claims, and ambiguous or coreference resolution issues. Such defects compound under iterative use, degrading retrieval fidelity and downstream task performance. We present \textbf{DeepRefine}, a general LLM-based reasoning model for \emph{agent-compiled knowledge refinement} that improves the quality of any pre-constructed knowledge bases with user queries to make it more suitable for the downstream tasks. DeepRefine performs multi-turn interactions with the knowledge base and conducts abductive diagnosis over interaction history, localizes likely defects, and executes targeted refinement actions for incremental knowledge base updates. To optimize refinement policies of DeepRefine without gold references, we introduce a Gain-Beyond-Draft (GBD) reward and train the reasoning process end-to-end via reinforcement learning. Extensive experiments demonstrate consistent downstream gains over strong baselines.
Code (0)
등록된 구현이 없습니다.
Tasks
Reinforcement LearningCoreference ResolutionSimilar Papers 제목 키워드 기반
ProgAgent:A Continual RL Agent with Progress-Aware Rewards
We present ProgAgent, a continual reinforcement learning (CRL) agent that unifies progress-aware reward learning with a high-throughput, JAX-native system architecture. Lifelong robotic learning grapples with catastrophi…
Reinforcement LearningContinual LearningCompiled Memory: Not More Information, but More Precise Instructions for Language Agents
Existing memory systems for language agents address memory management: how to retrieve and page more information within a context budget. We address a complementary problem -- memory utility: what experience is worth kee…
Transfer Learning for Automated Test Case Prioritization Using XCSF
With the rise of test automation, companies start to rely on large amounts of test cases. However, there are situations where it is unfeasible to perform every test case as only a limited amount of time is available. Und…
reinforcement-learningReinforcement Learning (RL)Transfer LearningConstraint-Guided Multi-Agent Decompilation for Executable Binary Recovery
Decompilation -- recovering source code from compiled binaries -- is essential for security analysis, malware reverse engineering, and legacy software maintenance. However, existing decompilers produce code that often fa…
Reinforcement Learning Integrated Agentic RAG for Software Test Cases Authoring
This paper introduces a framework that integrates reinforcement learning (RL) with autonomous agents to enable continuous improvement in the automated process of software test cases authoring from business requirement do…
Reinforcement Learning