A hybrid recommendation system for researchgate academic social network

Due to the exponentially increasing number of scholarly papers, academic information overload has caused significant difficulties for researchers. Papers recommendation is one of the methods proposed to help readers find interesting papers related to their work. According to a recent study by Nature journal, ResearchGate, with more than 15 million active members, is considered today as the largest academic social network. This platform offers scientists a multitude of functions, including sharing publications and projects, asking and answering questions, discussing research, establishing professional links, and making research visible. In this work, we propose a hybrid scientific paper recommendation system to help different actors on the ResearchGate professional network to carry out the various research and recruitment tasks. The proposed system combines three recommendation approaches, demographic-based, relationship network-based, and content-based techniques; and uses a dataset built by the authors. A data extraction API has been developed to build the experimental dataset containing most of the data types shared publicly on ResearchGate. Our dataset consists of more than 130000 researcher profiles, 15000 publications, and 1700 projects. The experiments carried out in this context, as well as the initial results, showed good performance, close to those obtained by the ResearchGate recommender system.