Designing a recommender system for Persian and English articles using the BERT language model, focusing on article abstract, title and keywords

Two factors of the rapid growth of various sciences in recent years and the growth and spread of the global Internet network have caused several thousand articles in various fields to be published on the Internet every day, and become available to everyone. This problem creates a challenge for the specialists of different sciences, because they have to look for their interests and scientific field in a large number of published articles so that they can always remain an expert. In this research, for the first time, we intend to introduce a recommender system that suggests suitable articles to Persian language users according to their interests. The purpose of this research is to provide basic research for this category of issues in Persian language, to be a starting point for other researchers. In this study, we want to propose articles based on the semantic similarity of the title, abstract and keywords of the article (with appropriate weighting) with the user's academic records. After the implementation and placement of the model as a pilot on the internet platform. A group of undergraduate students evaluated the model's computer software, and the overall accuracy of the system was evaluated at 79%.