Simulating Research Cooperation via LLM Agents with Social Theory-Driven Prompting

ACL ARR 2025 May Submission7740 Authors

Abstract: Research cooperation plays now more crucial role than ever in facilitating discoveries and enhancing innovation. Exploring how researchers select potential collaborators becomes increasingly important for understand and predict the research cooperation relationship. However, existing theory-driven empirical test studies fail to capture the overall context of research cooperation beyond numeric variables. In this paper, combing the advantages of Large Language Model (LLM) agent in context understanding and theory analysis in interpretability, we introduce a social theory LLM agent (ST-Agent) framework for simulating research cooperation. Research cooperation-oriented social theories (Social Exchange Theory and Cost-Benefit Theory)-guided prompts are proposed for LLM agent to first generate a theoretical analysis report on the motivation of research cooperation and then make cooperation potential prediction based on the report. Experimental results demonstrate the effectiveness of our method.

Paper Type: Long

Research Area: Computational Social Science and Cultural Analytics

Research Area Keywords: Computational Social Science and Cultural Analytics, Interpretability and Analysis of Models for NLP

Contribution Types: Model analysis & interpretability, Data analysis

Languages Studied: English, Chinese

Submission Number: 7740