Saved in:
Bibliographic Details
Main Authors: Liu, Chang, Li, Xiaoguang, Shang, Lifeng, Jiang, Xin, Liu, Qun, Lam, Edmund Y., Wong, Ngai
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.08148
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911498847125504
author Liu, Chang
Li, Xiaoguang
Shang, Lifeng
Jiang, Xin
Liu, Qun
Lam, Edmund Y.
Wong, Ngai
author_facet Liu, Chang
Li, Xiaoguang
Shang, Lifeng
Jiang, Xin
Liu, Qun
Lam, Edmund Y.
Wong, Ngai
contents Recently, large language models (LLMs) have gained much attention for the emergence of human-comparable capabilities and huge potential. However, for open-domain implicit question-answering problems, LLMs may not be the ultimate solution due to the reasons of: 1) uncovered or out-of-date domain knowledge, 2) one-shot generation and hence restricted comprehensiveness. To this end, this work proposes a gradual knowledge excavation framework for open-domain complex question answering, where LLMs iteratively and actively acquire external information, and then reason based on acquired historical knowledge. Specifically, during each step of the solving process, the model selects an action to execute, such as querying external knowledge or performing a single logical reasoning step, to gradually progress toward a final answer. Our method can effectively leverage plug-and-play external knowledge and dynamically adjust the strategy for solving complex questions. Evaluated on the StrategyQA dataset, our method achieves 78.17% accuracy with less than 6% parameters of its competitors, setting new SOTA for ~10B-scale LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2603_08148
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Gradually Excavating External Knowledge for Implicit Complex Question Answering
Liu, Chang
Li, Xiaoguang
Shang, Lifeng
Jiang, Xin
Liu, Qun
Lam, Edmund Y.
Wong, Ngai
Computation and Language
Artificial Intelligence
Recently, large language models (LLMs) have gained much attention for the emergence of human-comparable capabilities and huge potential. However, for open-domain implicit question-answering problems, LLMs may not be the ultimate solution due to the reasons of: 1) uncovered or out-of-date domain knowledge, 2) one-shot generation and hence restricted comprehensiveness. To this end, this work proposes a gradual knowledge excavation framework for open-domain complex question answering, where LLMs iteratively and actively acquire external information, and then reason based on acquired historical knowledge. Specifically, during each step of the solving process, the model selects an action to execute, such as querying external knowledge or performing a single logical reasoning step, to gradually progress toward a final answer. Our method can effectively leverage plug-and-play external knowledge and dynamically adjust the strategy for solving complex questions. Evaluated on the StrategyQA dataset, our method achieves 78.17% accuracy with less than 6% parameters of its competitors, setting new SOTA for ~10B-scale LLMs.
title Gradually Excavating External Knowledge for Implicit Complex Question Answering
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2603.08148