Open Science Behavior of AI Industry: Collaboration Patterns and Topics from the Perspective of Cross-Institutional Authors

In this study, the open science behaviour of industry is analyzed in terms of publications co-authored by at least one author from industry. Firstly, authors are classified into five types according to affiliations, and collaboration network is built to investigate the role of different types of authors. Then, topic evolution of papers related to industry is studied using Hierarchical Dirichelet Process and topic mapping algorithm. Artificial intelligence research area is taken as an example. Results show cross-university-industry type authors play an important bridge role in the collaboration network. And research topics that industries are concerned about are revealed and analyzed. This research could provide reference for formulating science and technology policy and promoting scientific research innovation in both universities and industries.