Profiles Matching in Social Networks Based on Semantic Similarities and Common Relationships
Ali Choumane, Zein Al Abidin Ibrahim, Bilal Chebaro · 2017
Social networks have experienced an explosion in both the number of users and shared data. Users of social network sites are constantly troubled by information overload as there are too many people to interact with and too much content to consult. So, helping users to find new relationships and relevant content is becoming a major challenge for these sites. The most popular social networks now offer tools to search content, people, pages, etc. but also recommend items of interest to the user. We cite as an example the famous phrase used by Facebook "People you may know" followed by a list of people who often belong to the user's environment (friends of friends), working or having worked in the same company, etc. This type of recommendation explores the links between users and few simple attributes (work company, address, ...) but neglects the published content. In this article, we propose a new persons' recommendation approach based on profiles matching by integrating both semantic similarities and common relationships between users. We conducted an experimental study, on real data from Twitter, that compares our approach with lexical and semantic matching methods.