一种基于调用序列网络的API推荐方法  

An API Recommendation Method Based on Call Sequence Network

作　　者：肖海涛 王鹏 包义祥 何鹏 XIAO Hai-tao;WANG Peng;BAO Yi-xiang;HE Peng(School of Computer and Information Engineering,Hubei University,Wuhan 430062,China)

机构地区：[1]湖北大学计算机与信息工程学院,湖北武汉430062

出　　处：《软件导刊》2018年第7期79-82,共4页Software Guide

基　　金：国家重点基础研究发展计划项目(2014CB340401);湖北省自然科学基金青年科学基金项目(2016CFB309)

摘　　要：随着计算机程序的日益复杂,代码自动补全功能需求越来越迫切。围绕软件编码过程中API调用问题进行探究,利用代码中API之间的调用序列,构建API关系网络模型,从服务推荐角度实现精准的API推荐,从而提高软件项目开发效率。实验结果表明,基于API序列关系网络模型推荐方法具有可行性,且在推荐列表长度较大的情况下方法更具优势,相比基准方法推荐精度可提高7.5%。在推荐过程中提供的API子序列越长,推荐结果越准确,但耗时明显增加。在子序列长度为5时,方法推荐精度与运行时间可达到相对适中的效果。With the complexity of computer program grows,the functional requirement of automatic code completion becomes more and more urgent.The paper explores the problem of API calls in software coding,using the call sequence between code of API and building API network model.It achieve accurate API recommendation from the perspective of service recommendation so as to improve the efficiency of software development projects.The experimental results verify the feasibility of the recommended method based on API sequence relation network models,and the method is more advantageous when the recomendation list is longer for the recommendation accuracy can be increased by 7.5% compared with the benchmark method.The recommended result is more accurate when the API sequence is longer in the process of recommendation.On the whole,when the subsequence length is 5,the recommendation accuracy and running time can achieve relatively modest results.

关 键 词：API推荐 服务计算 复杂网络 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]