人机智能协商是指通过智能体与人类对手进行协商，达到解决冲突、减少资源投入和优化协商结果的目的。近年来人机智能协商在电子商务和协商技能培训等方面具有重大的市场需求，已成为人工智能的一个重要研究课题。然而现有的人机智能协商研究忽略了协商启动机制方面的研究，导致在启动阶段就存在用户偏好信息获取困难、多边参与者匹配不合理等问题。这些问题的存在使得目前的人机智能协商系统尚不能在现实场景中得到有效应用。本项目通过研究协商启动机制的三个关键技术解决以上问题。在技术上，首先构建基于目标约束的偏好表示方法和用户偏好信息推理模型，其次利用基于自适应神经网络的模糊推理系统和信任网络理论设计多边协商者匹配方法，最后基于情感分析技术构建人类对手协商模型分类器。在应用上，通过整合启动机制相关技术来优化人机智能协商系统。本项目将有助于完善人机智能协商理论，对人机智能协商在复杂情景下的应用产生积极影响。

Human-agent automated negotiation refers to employing agents to negotiate with human opponents, which aims to solve conflicts, reduce resource investment and optimize negotiation results. In recent years, human-agent automated negotiation has great demand in e-commerce and negotiation skill training market, which has become an important research topic of artificial intelligence. However, the existing work of human-agent automated negotiation ignores the research on the start-up mechanism of negotiation, which leads to the problems in the start-up stage of negotiation, such as the difficulty of obtaining the user preference information and the unreasonable matching of negotiators in multilateral negotiation. These problems make the current human-agent automated negotiation systems cannot be effectively applied in real scenes. This project solves the above problems by studying three key technologies of the negotiation start-up mechanism. In the technology part this project will firstly use the purpose constraint-based preference representation method to explore the human user preference simulation and user preference information reasoning model. Secondly, the multilateral negotiator matching method will be designed through the fuzzy reasoning system based on adaptive neural network and trust network theory. Finally, the classifier of human opponent negotiation model will be constructed based on sentiment analysis technologies. In the application part this project will integrate the technologies of the start-up mechanism to optimize the human-agent automated negotiation system. The research results of this project will improve the theory of human-agent automated negotiation and have a positive impact on the application of human-agent automated negotiation in complex scenarios.