TY - GEN
T1 - Open Source Software Supply Chain Recommendation Based on Heterogeneous Information Network
AU - Lin, Hai Ming
AU - Liang, Guanyu
AU - Wu, Yanjun
AU - Wu, Bin
AU - Tian, Chunqi
AU - Wang, Wei
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - In the GitHub open-source collaborative development scenario, each entity type and the link relationship between them have natural heterogeneous attributes. In order to improve the accuracy of project recommendation, it is necessary to effectively integrate this multi-source information. Therefore, for the project recommendation scenario, this paper defines an open source weighted heterogeneous information network to represent the different entity types and link relationships in the GitHub open source collaborative development scenario, and effectively model the complex interaction among developers, projects and other entities. Using the weighted heterogeneous information network embedding method, extract and use the rich structural and semantic information in the weighted heterogeneous open source information network to learn the node representation of developers and projects, and fuse the personalized nonlinear fusion function into the matrix decomposition model for open source project recommendation. Finally, this paper makes a large number of comparative experiments based on the real GitHub open data set, and compares it with other project recommendation methods to verify the effectiveness of our proposed open source project recommendation model. At the same time, it also explores the impact of different metapaths on the effect of project recommendation. The experimental results show that the recommendation method based on heterogeneous information network can effectively improve the recommendation quality.
AB - In the GitHub open-source collaborative development scenario, each entity type and the link relationship between them have natural heterogeneous attributes. In order to improve the accuracy of project recommendation, it is necessary to effectively integrate this multi-source information. Therefore, for the project recommendation scenario, this paper defines an open source weighted heterogeneous information network to represent the different entity types and link relationships in the GitHub open source collaborative development scenario, and effectively model the complex interaction among developers, projects and other entities. Using the weighted heterogeneous information network embedding method, extract and use the rich structural and semantic information in the weighted heterogeneous open source information network to learn the node representation of developers and projects, and fuse the personalized nonlinear fusion function into the matrix decomposition model for open source project recommendation. Finally, this paper makes a large number of comparative experiments based on the real GitHub open data set, and compares it with other project recommendation methods to verify the effectiveness of our proposed open source project recommendation model. At the same time, it also explores the impact of different metapaths on the effect of project recommendation. The experimental results show that the recommendation method based on heterogeneous information network can effectively improve the recommendation quality.
KW - GitHub project recommendation
KW - Heterogeneous Information Network
KW - Open source
UR - https://www.scopus.com/pages/publications/85161225561
U2 - 10.1007/978-3-031-31180-2_5
DO - 10.1007/978-3-031-31180-2_5
M3 - 会议稿件
AN - SCOPUS:85161225561
SN - 9783031311796
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 70
EP - 86
BT - Benchmarking, Measuring, and Optimizing - 14th Bench Council International Symposium, Bench 2022, Revised Selected Papers
A2 - Gainaru, Ana
A2 - Zhang, Ce
A2 - Luo, Chunjie
PB - Springer Science and Business Media Deutschland GmbH
T2 - 14th International Symposium on Benchmarking, Measuring, and Optimization, Bench 2022
Y2 - 7 November 2022 through 9 November 2022
ER -