Enhancing a User Matchmaking Algorithm using Personalized PageRank
Santipong Thaiprayoon, Herwig Unger · 2023
With the increasing number of users in online communities and social networking platforms, it is becoming more difficult for users to meet and connect with individuals who share similar opinions or interests. The paper proposes a user matchmaking algorithm based on personalized PageRank to provide potential friends to individual users. A set of user profiles is transformed into a graph model for efficiently discovering meaningful connections and influential users. The semantic relationship between two user profiles is then estimated using word and sentence embeddings. By incorporating both embedding models and personalized graph analytics, the proposed algorithm can capture complex semantic information and high-order user relationships, making the matchmaking process more accurate. Experiments conducted on a simulated user profile dataset show that the proposed algorithm consistently outperforms existing state-of-the-art methods in terms of the F1 score and mean average precision metrics.