基于多源数据融合的Java代码知识图谱构建方法研究  

A Construction Method of Java Code Knowledge Graph based on multi-source data fusion

作　　者：苏佳 苏小红[1] 王甜甜[1] SU Jia;SU Xiaohong;WANG Tiantian(School of Computer Science and Technology,Harbin Institute of Technology,Harbin 150001,China)

机构地区：[1]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《智能计算机与应用》2020年第5期9-13,共5页Intelligent Computer and Applications

基　　金：国家自然科学基金项目(61672191)。

摘　　要：开发人员在软件开发过程中有学习和检索来自不同来源不同种类的代码相关知识的需要,代码知识图谱不仅能够对知识资源进行描述,还可以分析和描绘知识间的联系,越来越受到重视。本文提出了一种Java代码知识图谱的构建方法,将提取的Java代码相关的多源知识融合,进行知识图谱构建,并对现有API实体识别方法进行了实验和比较。Developers need to learn and retrieve different kinds of code-related knowledge from different sources in the software development process.Code knowledge graph can not only describe knowledge resources,but also analyze and display the relationship between knowledge,which has been paid more and more attention.In this paper,a method of constructing knowledge graph of Java code is proposed.Multi-source knowledge related to Java code is extracted and fused to construct knowledge graph,and existing API entity recognition methods are tested and compared.

关 键 词：代码知识图谱 知识融合 实体识别 知识图谱构建 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]