PD$^3$: A Project Duplication Detection Framework via Adapted Multi-Agent Debate

2025-05-23

AI Summary

To address shallow semantic understanding, poor interpretability, and the absence of expert cognitive modeling in duplicate detection for power-sector projects, this paper proposes a high-semantic-matching framework based on multi-agent collaborative debate. We introduce a novel adaptive multi-agent debate paradigm that integrates fair-competition-based semantic retrieval with a dual-modal (text + metrics) feedback generation mechanism, unifying qualitative analysis and quantitative evaluation. By incorporating domain knowledge enhancement, large language model (LLM)-enabled collaborative reasoning, and expert debate simulation, our method achieves performance gains of 7.43% and 8.00% on two downstream tasks across 800+ real-world power projects spanning over 20 subdomains. The system has been deployed in the review platform “Review Dingdang,” supporting preliminary screening for over 100 new projects and yielding cost savings of USD 5.73 million.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Multiagent Systems: Teamwork

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Project duplication detection is critical for project quality assessment, as it improves resource utilization efficiency by preventing investing in newly proposed project that have already been studied. It requires the ability to understand high-level semantics and generate constructive and valuable feedback. Existing detection methods rely on basic word- or sentence-level comparison or solely apply large language models, lacking valuable insights for experts and in-depth comprehension of project content and review criteria. To tackle this issue, we propose PD$^3$, a Project Duplication Detection framework via adapted multi-agent Debate. Inspired by real-world expert debates, it employs a fair competition format to guide multi-agent debate to retrieve relevant projects. For feedback, it incorporates both qualitative and quantitative analysis to improve its practicality. Over 800 real-world power project data spanning more than 20 specialized fields are used to evaluate the framework, demonstrating that our method outperforms existing approaches by 7.43% and 8.00% in two downstream tasks. Furthermore, we establish an online platform, Review Dingdang, to assist power experts, saving 5.73 million USD in initial detection on more than 100 newly proposed projects.

Problem

• Detects project duplication to improve resource utilization efficiency
• Overcomes limitations of basic word/sentence-level comparison methods
• Provides qualitative and quantitative feedback for expert review

Innovation

• Adapted multi-agent debate framework
• Qualitative and quantitative feedback analysis
• Online platform for expert assistance