Skip to main navigation Skip to search Skip to main content

TPE: Towards Better Compositional Reasoning over Cognitive Tools via Multi-persona Collaboration

  • Hongru Wang
  • , Huimin Wang
  • , Lingzhi Wang
  • , Minda Hu
  • , Rui Wang
  • , Boyang Xue
  • , Yongfeng Huang
  • , Kam Fai Wong*
  • *Corresponding author for this work
  • Chinese University of Hong Kong

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Previous works in tool learning mainly focus on the external function tools, such as models and APIs, while overlooking the existence of cognitive tools inside the cognitive/thinking processing of human mind. In this paper, we expand the scope of tools, centering on cognitive tools within the context of dialogue systems, such as multiple sources and psychological or tutoring strategies being dynamically applied in a single turn to compose helpful responses. To further enhance the reasoning and planning capability of LLMs with these cognitive tools, we introduce a multi-persona collaboration framework: Think-Plan-Execute (TPE). This framework decouples the response generation process into three distinct roles: Thinker, Planner, and Executor. Specifically, the Thinker analyzes the internal status exhibited in the dialogue context, such as user emotions and preferences, to formulate a global guideline. The Planner then generates executable plans to call different cognitive tools (e.g., sources or strategies), while the Executor compiles all intermediate results into a coherent response. Experimental results on three public datasets (FoCus, CIMA, and PsyQA) demonstrate the effectiveness and robustness of TPE, revealing the promising potential of cognitive tools as a substantial expansion of current toolset.

Original languageEnglish
Title of host publicationNatural Language Processing and Chinese Computing - 13th National CCF Conference, NLPCC 2024, Proceedings
EditorsDerek F. Wong, Zhongyu Wei, Muyun Yang
PublisherSpringer Science and Business Media Deutschland GmbH
Pages281-294
Number of pages14
ISBN (Print)9789819794331
DOIs
StatePublished - 2025
Externally publishedYes
Event13th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2024 - Hangzhou, China
Duration: 1 Nov 20243 Nov 2024

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume15360 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference13th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2024
Country/TerritoryChina
CityHangzhou
Period1/11/243/11/24

Keywords

  • Dialogue System
  • Large Language Model
  • Tool Learning

Fingerprint

Dive into the research topics of 'TPE: Towards Better Compositional Reasoning over Cognitive Tools via Multi-persona Collaboration'. Together they form a unique fingerprint.

Cite this