IntrAgent: An LLM Agent for Content-Grounded Information Retrieval through Literature Review
The paper introduces IntrAgent, an LLM-based agent designed to automate content-grounded information retrieval through literature review by mimicking human reading behaviors. It proposes a new task called IntraView and a two-stage process involving section ranking and iterative reading to extract precise, contextually grounded answers. Evaluated on the IntraBench benchmark across five STEM domains, IntrAgent outperforms existing RAG and research-agent baselines by an average of 13.2% in cross-domain accuracy. The work aims to improve accuracy and fidelity in scientific information retrieval.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Information Retrieval arXiv:2604.22861 (cs) [Submitted on 23 Apr 2026] Title:IntrAgent: An LLM Agent for Content-Grounded Information Retrieval through Literature Review Authors:Fengbo Ma, Zixin Rao, Xiaoting Li, Zhetao Chen, Hongyue Sun, Yiping Zhao, Xianyan Chen, Zhen Xiang View a PDF of the paper titled IntrAgent: An LLM Agent for Content-Grounded Information Retrieval through Literature Review, by Fengbo Ma and 7 other authors View PDF Abstract:Scientific research relies on accurate information retrieval from literature to support analytical decisions.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI