随着Siri及Waston的出现，交互式问答技术越来越引起人们的关注。问答机器人以自动的方式对问题进行理解和回答，在企业在线客服、教育、政务咨询等方面有着广泛的应用。本项目主要开展以下几个的方面的研究：1）问题理解技术。针对交互式问答中，问句中存在子问题、大量的省略现象以及基于匹配的检索模型无法很好解决的问句检索问题，本项目开展了问句拆分、对话中省略句判别及恢复及词项赋权等技术的研究。2）相似、相关问题推荐技术。当用户无法很好给出问题的描述时，相似、相关问题的推荐在交互式问答系统中就显得格外重要。本项目在机器学习和自然语言处理方法的基础上，开展了相似、相关问题推荐技术的研究。本项目的研究内容对于相关研究提供了重要的理论基础；词项赋权、对话中省略恢复等核心技术的研究，对于推动交互式问答技术有着重要的价值。

With the emerging of Siri and Waston, interactive question answering (iQA) technique becomes more and more attractive. Question answering (QA) robot based on automatic question understanding and answering is widely applied to online customers of corporations, consultations of education and government affairs, etc. This study mainly focuses on the following techniques: 1) Question understanding technique. For the existence of sub-questions, the amount of ellipsis and the intrinsic problems of the matching based retrieval models on question retrieval task, the techniques of question decomposition, ellipsis sentence identification and recovery on dialogue and question term weighting are proposed in this study. 2) Similar and relevant question recommendation. When question descriptions are absent, the similar and relevant question recommendation techniques are integrant in iQA systems. In this study, we will perform the similar and relevant question recommendation work based on the machine learning and natural language processing approaches. The substantial theories proposed by this study are basis for prevalent research. Furthermore, the development of iQA can be boosted by the key techniques of this study, such as term weighting, ellipsis recovery on dialogue, etc.