微博实时的信息分享使得基于微博内容开展用户位置研究变成一个重要研究内容，这类技术的发展对个性化地理服务、广告营销和智慧城市等领域提供了科学技术层面的支持。现有研究工作主要依赖于用户的主动地理信息分享，由于其位置分析精细度较低并且用户行为意图信息不明，这些技术很难支撑上述应用。本课题旨在通过分析用户微博内容，依次研究微博地理兴趣点（POI）抽取算法， POI地理位置推断算法和用户关于POI的行为意图分析算法，实现细粒度地获知用户的地理位置、活动计划和内容的目标。项目将采用文本挖掘与自然语言处理相关技术理论，重点研究:（1）POI词条的语义语境特征的抽取、转换和表征方法;（2）基于空间行为模式与空间主题分布的POI位置推断方法；（3）用户关于POI的时间趋势特征抽取与度量方法。本项研究对于推动文本处理技术的进一步发展以及满足应用领域对用户地理信息分析的需求，具有重要的科学意义和应用价值。

Given the real-time information sharing in microblogs, user location identification based on the content of microblogs has become an important research area. This class of techniques provides scientific support for personalized location services, advertisement and smart city, and so on. Existing studies mainly reply on the explicit location sharing of the users, which cannot provides fine-grained location estimation, or the extracted user behavior intent is unknown for the above applications. In this project, based on the content of microblogs, our endeavor is devoted to the development of point-of-interest (POI) extraction, POI location identification and POI-specific user behavior intent analysis, aiming at inferring user’s location, activity schedule and content in a fine-grained scale. Following the text mining and natural language processing methodology, we intensively address the following aspects: (1) the extraction, transformation and representation of the semantic and contextual features regarding POI; (2) POI location estimation approach based on both the spatial behavior pattern and spatial topic distribution; (3) the extraction and measurement of the temporal awareness features regarding POI. This project will make significant contribution to the development of text processing techniques and fulfill the requirements on the user geographical data analysis from industry.