中文摘要：

传统搜索引擎需要用户从返回网页中提炼有用知识;社交网络搜索根据人物的社会关系、共同爱好,提供人物和兴趣间的关系等方面的搜索结果。当前,社交网络搜索主要存在2个问题：不能从语义上理解用户查询词;仅局限于人物、兴趣搜索,限制了查询范围。为解决微博搜索中存在的一些问题,并主动返回更多知识,基于微博这一社交网络的重要平台,研究微博社区知识图谱构建方法,重点提出5方面的研究：微博社区中概念提取,其概念包括人物、事物、地点、事件和话题等5种类型;微博社区概念间的关系提取,其关系包括上述5种概念间的组合关系;知识图谱是带有语义的网络图谱,将概念作为顶点并将概念间关系作为边,研究知识图谱的构建方法;分析微博社区知识图谱,包括构建效果、演化特征、应用效果分析;研发基于微博知识图谱的应用系统等内容。

英文摘要：

Search engine only returns the Web page set for the user queries, it needs the user refine useful knowledge from it; Social Network Search （SNS） directly provides people and their interest to users by using characters＇ social relations and common hobbies. However, the SNS mainly exists two unresolved problems. On the one hand, the SNS can＇ t semantically understand user queries submitted by users. On the other hand, the SNS only provides people search and interest search, and confines query domains for users. Microblog has become an important platform for social network. To address these problems of information retrieval about microblog and provide more knowledge for user queries, this project researches knowledge graph construction and analysis based on the microblog community. The project focuses on five contents. （ 1 ） It researches concept extractions for the microblog community, and concepts have five types including people, things, locations, events and topics. （2）It researches relationships extractions for the microblog community. The relationships among concepts include collection types formed by combining two arbitrary types above concepts. （3） It researches knowledge graph construction, and the knowledge graph is a semantic network graph which takes concepts and relationships respectively as vertices and edges. （4）It researches knowledge graph analysis. It includes construction effect analysis, evolution characteristics and rules analysis and application effect analysis. （5）It researches the application interface and system based the knowledge graph.