中国邮电高校学报(英文) ›› 2020, Vol. 27 ›› Issue (1): 10-25.doi: 10.19682/j.cnki.1005-8885.2020.0006

• Computer Applied Technology • 上一篇    下一篇

IDL: Evaluating software quality based on PageRank algorithm

zhouguoqiang1,FAN Yi1,ZHANG Shuai1,WANG Yi-lun1,DAI Gui-lan2   

  1. 1. School of computer science, nanjing university of posts and telecommunications
    2. Institute of information technology,Tsinghua University, Beijing 100084,c
  • 收稿日期:2019-09-04 修回日期:2019-12-16 出版日期:2020-02-28 发布日期:2020-02-28
  • 通讯作者: DAI Gui-lan E-mail:daigl@tsinghua.edu.c

IDL: Evaluating software quality based on PageRank algorithm

  • Received:2019-09-04 Revised:2019-12-16 Online:2020-02-28 Published:2020-02-28
  • Contact: dai guilan E-mail:daigl@tsinghua.edu.c

摘要: In the traditional method, the software quality is measured by various metrics of the software, such as decoupling level (DL), which can be used to predict software defect. However, DL, which treats all the ?les equally, has not taken file importance into consideration. Therefore, a novel software quality metric, named as improved decoupling level (IDL), based on the importance of documents was proposed. First, the PageRank algorithm was used to calculate the importance of ?les to obtain the weights of the dependencies, and then defect prediction models was established by combining the software scale, dependencies, scores and software defects to assess the software quality. Compared to most existing module-based software quality evaluation methods, IDL has similar or even superior performance in the prediction of software quality. The results indicate that IDL measures the importance of each ?le in the software more accurately by combining the PageRank algorithm in DL, which indirectly re?ects the quality of software by predicting the bug information in software and improves the accuracy of prediction result of software bug information.

关键词: decoupling level (DL) , PageRank , improved decoupling level (IDL) , software architecture, software matrix , software quality

Abstract: In the traditional method, the software quality is measured by various metrics of the software, such as decoupling level (DL), which can be used to predict software defect. However, DL, which treats all the ?les equally, has not taken file importance into consideration. Therefore, a novel software quality metric, named as improved decoupling level (IDL), based on the importance of documents was proposed. First, the PageRank algorithm was used to calculate the importance of ?les to obtain the weights of the dependencies, and then defect prediction models was established by combining the software scale, dependencies, scores and software defects to assess the software quality. Compared to most existing module-based software quality evaluation methods, IDL has similar or even superior performance in the prediction of software quality. The results indicate that IDL measures the importance of each ?le in the software more accurately by combining the PageRank algorithm in DL, which indirectly re?ects the quality of software by predicting the bug information in software and improves the accuracy of prediction result of software bug information.

Key words: decoupling level (DL) , PageRank , improved decoupling level (IDL) , software architecture, software matrix , software quality