面向大规模知识图谱的查询处理关键技术研究

项目名称： 面向大规模知识图谱的查询处理关键技术研究

项目编号： No.61472085

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 肖仰华

作者单位： 复旦大学

项目金额： 83万元

中文摘要： 近年来，随着语义理解需求的日益增长，知识图谱日益成为一种重要的表达各类实体、概念及其之间语义关系的新型知识表示形态。同时，各类基于语义理解的实际应用对面向知识图谱的查询处理提出了需求。然而知识图谱的海量规模、异构信息、多模形态和复杂结构对当前的图查询处理技术提出了全新的挑战。针对这些挑战，本项目拟围绕基于知识图谱的语义应用，建立知识图谱的表示模型，揭示知识图谱的结构特征，研究大规模知识图谱的查询处理关键技术，突破大图分布式数据组织和复杂知识图谱语义查询处理关键技术，实现高效的大规模知识图谱查询处理和数据组织方法。本项目对于进一步提升图数据管理的研究水平、阐释基于知识图谱的语义理解机制具有重要的学术意义；对于支撑现实应用中的语义查询具有较高的实际应用价值。

中文关键词： 知识图谱；图数据；查询处理

英文摘要： With the increase of the requirement for semantic understanding, we have witnessed the emergence of many big knowledge graphs, which contain entities or concepts as the vertices and the semantic relationships as the edges. Meanwhile, many applications based on semantics understanding are demanding high quality query answering on knowledge graphs. However, there still exist many great challenges. In general, a typical knowledge graph is big, and it usually contains heterogeneous information, can be modeled by different models, and has a complex structure. To overcome these challenges, we propose this project to study the core query answering techniques on knowledge graphs, by employing the structural information of knowledge graph and summarizing the query model from real semantic based applications. We will investigate three key problems: the model of knowledge graphs, query answering algorithms and data organization techniques. We aim to proposing a collection of fundamental theories and efficient methods for query answering on knowledge graphs and overcome the challenge posed by the distributed big graph processing. The results of this project are good complements to the existing graph data processing techniques. The results are also helpful for the understanding the psychology mechanism of semantics on knowledge graphs. The proposed method can also find may valuable real applications.

英文关键词： Knowledge Graph;Graph Data;Query Processing