期刊论文

Problem Solver Recommendation Algorithm Based on Implicit Behavior

中文

小型微型计算机系统

1000-1220

2019

40

3

606-611

10.3969/j.issn.1000-1220.2019.03.026

国家社会科学基金一般项目（17BTQ061）资助；

本校为第一机构

从事软件开发工作的开发者经常会通过技术交流社区提出问题,给问题推荐合适的解决者是解决开发者遇到困难问题时的可行途径.目前的解决方案主要是基于开发者的显性行为(如:回复行为)而极少关注开发者的隐性行为(如:收藏行为),使得方案不能有效地发现潜在的问题解决者.为此,本文提出了一种基于隐性行为的问题解决者推荐算法,通过计算开发者的基于标签的隐性行为变量、解决问题的倾向性变量,并结合能力变量使用贝叶斯多变量回归得到开发者得分,排序后推荐问题解决者.实验过程基于互联网问答社区StackOverflow的数据,实验结果证明本文的方法具有良好的效果.

Developers engaged in software development work often raise questions through technical communication communities,and recommending appropriate solver for problems is a feasible way to solve developers' problems. The current solution is mainly based on the developer's explicit behavior (eg,reply behavior) and rarely on the developer's implicit behavior (eg,collection behavior),making the solution unable to effectively find potential problem solvers. For this reason,this paper proposes a problem-solver recommendation algorithm based on implicit behaviors,which uses the label-based implicit behav...