Developer Recommendation For Open Source Communities Based On Deep Learning And Hybrid Clustering

In recent years,with the advancement of Internet technology,open source software develop-ment methods for open source communities have developed rapidly with low-cost,high-quality development features.However,with the influx of software requirements and projects,huge competitive pressure makes it difficult for open source projects to quickly attract enough de-velopers to jointly promote the development of software projects.However,with the influx of software requirements and projects,huge competitive pressure makes it difficult for open source projects to quickly attract enough developers to jointly promote the development of software projects.In order to solve the above problems,this paper proposes a developer recommendation method for open source communities based on deep learning and hybrid clustering.Specifically,the main work and innovation of this paper are reflected in the following aspects:(1)In order to analyze developers from multiple perspectives,this article proposes a multi-index developer modeling method based on four dimensions of activity,influence,contribution,and development preference.(2)In order to recommend multiple categories of developers to software projects,it is neces-sary to predict the categories of developers who may participate in the project.This paper first proposes a developer classification algorithm based on hybrid clustering to classify de-velopers participating in the project.Then,based on the project document information,a developer category prediction algorithm based on convolutional neural network is proposed to predict the category of the developer participating in the project.(3)For extracting complex features from the original data for accurate developer recommen-dation,this article proposes a developer recommendation method based on the developer contribution relationship network and deep neural network.The developer contribution re-lationship network is used to calculate the developer's affected degree.The deep neural network predicts the possibility of developers participating in the project by synthesizing the project text,developer characteristics and developer's affected degree.After predict-ing the possibility of developers participating in the project,rank the developers in each developer category participating in the project.Finally,combine developer categories to recommend multi-category developers for software projects.(4)In order to verify the effectiveness of the method proposed in this paper,this research crawled and collected the developer and project information on Github to construct the ex-perimental data set of this article.Through a series of experiments,the proposed method proposed in this paper is compared with existing recommended algorithms.Experimental results proved that the method proposed in this paper had better recommendation ability than other methods.