Open Access

ARTICLE

HSPM: A Better Model to Effectively Preventing Open-Source Projects from Dying

by Zhifang Liao¹, Fangying Fu¹, Yiqi Zhao¹, Sui Tan^2,3,*, Zhiwu Yu^2,3, Yan Zhang⁴

1 School of Computer Science and Engineering, Central South University, Changsha, 410075, China
2 National Engineering Research Center of High-speed Railway Construction Technology, Changsha, 410075, China
3 School of Civil Engineering, Central South University, Changsha, 410075, China
4 School of Engineering and Built Environment, Glasgow Caledonian University, Glasgow, G4 OBA, UK

* Corresponding Author: Sui Tan. Email:

Computer Systems Science and Engineering 2023, 47(1), 431-452. https://doi.org/10.32604/csse.2023.038087

Received 26 November 2022; Accepted 03 March 2023; Issue published 26 May 2023

Abstract

With the rapid development of Open-Source (OS), more and more software projects are maintained and developed in the form of OS. These Open-Source projects depend on and influence each other, gradually forming a huge OS project network, namely an Open-Source Software ECOsystem (OSSECO). Unfortunately, not all OS projects in the open-source ecosystem can be healthy and stable in the long term, and more projects will go from active to inactive and gradually die. In a tightly connected ecosystem, the death of one project can potentially cause the collapse of the entire ecosystem network. How can we effectively prevent such situations from happening? In this paper, we first identify the basic project characteristics that affect the survival of OS projects at both project and ecosystem levels through the proportional hazards model. Then, we utilize graph convolutional networks based on the ecosystem network to extract the ecosystem environment characteristics of OS projects. Finally, we fuse basic project characteristics and environmental project characteristics and construct a Hybrid Structured Prediction Model (HSPM) to predict the OS project survival state. The experimental results show that HSPM significantly improved compared to the traditional prediction model. Our work can substantially assist OS project managers in maintaining their projects’ health. It can also provide an essential reference for developers when choosing the right open-source project for their production activities.

---

Keywords

---