MACHINE LEARNING FOR AUTOMATIC LABELLING OF USERS OF AN ACADEMIC SOCIAL NETWORK

Abstract

Social networks have become relevant in the Internet due to the great variety of Web sites that use the concept. Its users form databases that provide an important way of sharing, organizing, finding content and making contacts. Thus, Scientia.Net is a social networking site that integrates information from various Internet services (forums, article repositories, websites, blogs and other social networks). Besides, the tool provides the user interaction (students, teachers and researchers) for academic purposes, based on their common interests. This paper presents an application developed to automatically group Scientia.Net users, showing the performance of various machine learning algorithms, offering to Scientia.Net a sorting mechanism that presents a list of other researchers to each user of the network, based on their common interests. With this, we intend to contribute to the interaction among users with similar profiles, allowing an improvement in the productivity of their research efforts. Furthermore, this paper proposes a model that uses a combination of supervised and unsupervised learning algorithms to create groups and identify users based on their relevant attributes.