中文摘要：

知识库中存储着大量关于真实世界中的实体信息及实体之间的关系,随着规模的不断增长,其应用也愈发广泛。另一方面,由于大量互联网用户通过关键词描述问题和查询意图,因此如何让知识库具备更好的关键词查询应答能力,成为了研究的热点。从中文知识库的构建入手,提出了一套完整的面向中文限定领域知识库的关键词检索框架,实现并改进了基于模板的关键词查询转换方法,提出了基于语义的知识库释义和实体索引方法,提高了关键词映射能力。同时在SPARQL转换过程中采用了缺失关系索引,提高了转换成功率,提升了能够处理的查询数量。同时在学术空间ScholarSpace上对该框架进行了系统实现,取得了良好的应用效果。

英文摘要：

Knowledge bases（KB） store large amount of structured information about the entities and their relationships.As the scale of KBs increased, their application also varied. On the other side, large amount of users describe their question or query intention by submitting keyword queries. Thus enabling KB to answer these keyword queries becomes of crucial importance. A framework from building a Chinese KB to answering keyword search over it was established. A novel approach based on query template to translate the keyword queries into structured queries was proposed. A semantic based paraphrase and index approach to improve the result of query term mapping and an absent predicate index to deal with the predicate absence during the query translation step was proposed. Significant improvement of the ability of translating keyword query to structured query was achieved. Finally the framework and approach was implemented in the ScholarSpace system and get a good performance.