徐健



【摘 要】
随着5G技术的日趋成熟,运营商相关研究人员面临着快速掌握5G相关知识的压力,然而5G知识内容种类繁多,知识面广,如何高效地从5G协议中查询到亟需学习的知识点是当前亟待解决的问题,为了解决这一问题,本文基于知识图谱及信息搜索方法构建5G协议知识库。由于网优人员搜索相关知识一般只需得到与关键字相关的知识信息,而不必通晓全文,本文采用专业领域知识库结合多种方法对信息进行抽取,建立术语的属性、基本关系以及文本结构的关系,同时采用Neo4j图数据库对构建成的三元组进行存储,极大地提高了搜索性能,本文将该知识图谱运用到5G协议知识智能检索中,并取得了很好的效果。
【关键词】知识图谱;Neo4j;搜索引擎;深度学习;知识抽取
[Abstract]
With the maturity of 5G technology, operator-related researchers face the pressure to quickly acquire 5G knowledge. However, there are various types of 5G knowledge content with a wide range, and how to efficiently extract the knowledge points from 5G protocol is an urgent issue to be solved. In order to solve it, this paper constructs a 5G protocol knowledge base using the methods of knowledge graph and information search. Since network optimization engineers usually need to get knowledge related to key words rather than being familiar with the full text when searching relevant knowledge, this paper adopts professional domain knowledge bases and combines multiple methods to extract information to establish term attributes, basic relationships and text structures. At the same time, the paper also uses the Neo4j graph database to store the construed triples, which greatly improves the search performance. This paper applies the knowledge graph to intelligent retrieval in 5G protocol knowledge and obtains a successful achievement.
[Key words]knowledge graph; Neo4j; search engine; deep learning; knowledge extraction
0 引言
随着网络领域人工智能技术的发展,自然语言处理技术得到了进一步的发展,由于网优领域知识的大规模、非结构化等特点,这使网优人员获取信息的难度加大,同时,知识图谱的快速发展,又给网优知识整理提供了可能性,知识图谱以其强大的语义处理能力和开放性组织能力,为网络领域的知识整理和自动化应用打下了根基,许多领域面临数据的不断增长所带来的许多挑战,因此可以利用知识图谱技术,针对不同的业务需求,实现通用领域和专用领域应用“遍地开花”的景象。
DBpedia是知识图谱中很典型的例子,是从维基百科的结构化数据中提取出来的知识图,这种提取的数据主要来源是维基百科信息框中的键值对,在一个众包过程中,提取信息框中的内容作为实体,而相应的键值作为属性,基于这些映射,可以提取知识图[1]。……