面向大规模知识图谱的弹性语义推理方法研究及应用

发布时间：2018-03-28 20:05

本文选题：知识图谱　切入点：语义推理　出处：《浙江大学》2017年博士论文

【摘要】：知识图谱可以看作是人类刻画和认知世界的一个载体,是人工智能的重要分支,它可以对现实世界的事物及其相互关系进行形式化地描述,以刻画和揭示实体之间的语义关系,进而模仿人类的思维方式进行逻辑推理。知识图谱的推理技术是知识图谱最核心的技术之一,表示由一个或几个已知的判断推理出未知结论的过程,广义的推理包含语义查询和语义推理,本文所指的推理是广义范围的推理技术。近年来,随着知识抽取技术不断发展以及数据来源的不断丰富,知识图谱的规模正在迅速增长,当前知识图谱默认已被用来代表各种海量知识图谱。伴随着知识图谱规模的增长,知识图谱呈现出结构复杂多样性、数据动态变化性以及查询实时响应性等多种特性和需求,面对知识图谱所呈现的新特征和挑战,传统的语义推理方法和工具缺乏相应的对策,这极大限制了知识图谱技术的进一步推广和发展,研究如何高效的在大规模知识图谱上实施弹性语义推理成为一个重要研究课题。本文从大规模知识图谱语义推理所面临的挑战出发,研究了面向大规模知识图谱弹性语义推理的若干实用性科学问题,研究内容主要包括以下几个方面:(ⅰ)针对大规模知识图谱复杂知识结构的挑战,研究基于OWL属性链的弹性语义关联推理方法。(ⅱ)针对大规模知识图谱查询实时性的挑战,研究基于分布式内存的弹性语义查询推理方法。(ⅲ)针对大规模知识图谱流式动态性的挑战,研究基于规则的弹性语义流推理方法。(ⅳ)基于上述的方法和经验,设计实施了两个实用的面向大规模知识图谱弹性语义推理系统:基于Spark的大规模知识图谱规则推理系统和领域相关的大规模中西医知识图谱关联发现系统。
[Abstract]:Knowledge map can be regarded as a carrier of human portrayal and cognition of the world and an important branch of artificial intelligence. It can describe things in the real world and their interrelations formally to depict and reveal the semantic relationship between entities. The reasoning technique of knowledge map is one of the core techniques of knowledge map, which represents the process of inferring an unknown conclusion from one or more known judgments. Generalized reasoning includes semantic query and semantic reasoning. The reasoning in this paper is a generalized reasoning technique. In recent years, with the development of knowledge extraction technology and the continuous enrichment of data sources, the scale of knowledge map is growing rapidly. At present, the knowledge map has been used to represent all kinds of mass knowledge atlas by default. With the increase of the scale of knowledge map, the knowledge map presents a variety of characteristics and requirements, such as complex structure diversity, data dynamic variability and query real-time response. In the face of the new characteristics and challenges of knowledge map, the traditional semantic reasoning methods and tools lack corresponding countermeasures, which greatly limit the further promotion and development of knowledge map technology. It has become an important research topic to study how to efficiently implement elastic semantic reasoning on large-scale knowledge atlas. This paper starts from the challenge of large-scale knowledge map semantic reasoning. In this paper, some practical scientific problems of flexible semantic reasoning for large-scale knowledge atlas are studied. The research contents mainly include the following aspects: (I) the challenge of complex knowledge structure in large-scale knowledge atlas. This paper studies the flexible semantic association reasoning method based on OWL attribute chain. (II) the challenge of real-time query based on large-scale knowledge atlas. This paper studies the flexible semantic query reasoning method based on distributed memory. (III) aiming at the challenge of large-scale knowledge map flow dynamics, this paper studies the rule-based elastic semantic flow reasoning method. (IV) based on the above methods and experiences, Two practical flexible semantic inference systems for large-scale knowledge atlas are designed and implemented: the large-scale knowledge atlas rule inference system based on Spark and the domain-related large-scale knowledge atlas association discovery system of traditional Chinese and Western medicine.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP18;TP391.3

【相似文献】