SSC-CF: Semantic similarity and clustering-based collaborative filtering for expert recommendation in community question answering websites

• Published in: International journal of information technology (Singapore. Online) Vol. 15; no. 8; pp. 4243 - 4257
• Main Authors: Paramasivam, Aarthi, Nirmala, S. Jaya
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.12.2023; Springer Nature B.V
• Subjects: Artificial Intelligence; Clustering; Collaboration; Computer Imaging; Computer Science; Factorization; Filtration; Image Processing and Computer Vision; Machine Learning; Original Research; Pattern Recognition and Graphics; Questions; Recommender systems; Semantics; Similarity; Software Engineering; Sparsity; Vision; Natural language processing; Community question answering; Sparsity; Scalability; Expert recommendation
• ISSN: 2511-2104; 2511-2112
• DOI: 10.1007/s41870-023-01458-6

Community question answering forums allow users to find knowledge on a topic of interest by asking questions and getting answers from experts. However, it can be challenging to find experts who are knowledgeable in a particular subject, especially when there are millions of questions and thousands of new queries every day.This paper proposes a novel expert recommendation system called Semantic Similarity and Clustering-based Collaborative Filtering (SSC-CF). SSC-CF addresses two key drawbacks of collaborative filtering: scalability and sparsity. Sparsity is addressed by using matrix factorization. In matrix factorization, latent features are identified to detect similarity and generate a prediction based on both the question and the user entities. Whereas a clustering method is employed to group users and questions with shared interests to address scalability. The recommendation system’s accuracy is further improved by incorporating semantic similarity. SSC-CF is evaluated on three Stack Exchange sites: gaming, physics, and scifi. The results clearly show that the proposed technique, SSC-CF, is effective in addressing both scalability and sparsity.