ARC: Active and Reflection-driven Context Management for Long-Horizon Information Seeking Agents

2026-01-17

AI Summary

This work addresses the performance degradation of large language models in long-horizon information retrieval tasks, commonly attributed to “context corruption” caused by accumulating irrelevant context. To mitigate this issue, the authors propose an active, reflection-driven context management framework that treats context as a dynamic reasoning state. The framework employs real-time monitoring and revision mechanisms to assess task relevance and actively reconstruct the working context. This approach uniquely models context management as an ongoing, reflective process during execution, moving beyond conventional static compression or passive summarization strategies. Evaluated on benchmarks such as BrowseComp-ZH using Qwen2.5-32B-Instruct, the method achieves up to an 11% absolute improvement in accuracy over passive compression baselines.

Technology Category

• Natural Language Processing: Summarization
• Data Mining & Knowledge Management: Data Compression
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language models are increasingly deployed as research agents for deep search and long-horizon information seeking, yet their performance often degrades as interaction histories grow. This degradation, known as context rot, reflects a failure to maintain coherent and task-relevant internal states over extended reasoning horizons. Existing approaches primarily manage context through raw accumulation or passive summarization, treating it as a static artifact and allowing early errors or misplaced emphasis to persist. Motivated by this perspective, we propose ARC, which is the first framework to systematically formulate context management as an active, reflection-driven process that treats context as a dynamic internal reasoning state during execution. ARC operationalizes this view through reflection-driven monitoring and revision, allowing agents to actively reorganize their working context when misalignment or degradation is detected. Experiments on challenging long-horizon information-seeking benchmarks show that ARC consistently outperforms passive context compression methods, achieving up to an 11% absolute improvement in accuracy on BrowseComp-ZH with Qwen2.5-32B-Instruct.

Problem

• context rot
• long-horizon information seeking
• context management
• reasoning coherence
• interactive agents

Innovation

• active context management
• reflection-driven reasoning
• context rot
• long-horizon information seeking
• dynamic context revision