考虑开发者兴趣迁移的开源项目推荐方法

Open Source Software Project Recommendation Considering Migrations of Developers’ Interests

社交化编码是当前软件开发生产的一个重要方式,其开发模式的灵活性与开放性吸引了大量的开发者.开发者通过参与开源项目能够提升自己的开发能力并在社区中形成自己的影响力.许多开源社区如GitHub上有大量的开源项目.开发者将花费大量的时间与精力去寻找自己感兴趣的项目.开源项目推荐引起了研究者的兴趣,然而,目前的方法中仅仅基于开发者过去参加过的项目的相似性进行项目推荐,没有对开发者的兴趣迁移进行考虑.针对这一问题,本文提出了一种基于项目主题迁移频繁模式挖掘的推荐算法.该方法结合了概率主题模型与顺序频繁模式挖掘,并考虑项目社交关联和流行度,从而为开发者提供个性化开源项目推荐.本文所提方法的召回率比传统的方法高出了10.9%,推荐效果显著提升.

Social coding is an important way of current software development and production.The flexibility and openness of its development model attract a large number of developers.Developers participate open source projects in order to improve their capabilities and impacts in the community.Many open source platforms such as GitHub host a large number of open source projects.Developers have to spend a lot of time and efforts to search the projects they are interested in.Therefore,open source project recommendation becomes a hot research topic.However,the approaches proposed so far are only based on the similarities with the historical projects a developer participated in and do not consider the interest migration of developers.In this paper a frequent project topic migration pattern mining based approach is proposed.It combines the method of probabilistic topic model and sequential frequent pattern mining,and also considers the social connections and popularities of projects,to provide developers with personalized recommendations for open source projects.The recall rate of the method proposed in this article is 10.9% higher than the traditional method,and the recommendation effect is significantly improved.