A trust and semantic based approach for social recommendation

With the rapid advancement of Internet, e-commerce websites and social networks, people prefer to receive recommendations from their social friends rather than strangers. Also, the exponential evolution and use of online social networks has resulted in generation of enormous amount of information ov...

Full description

Saved in:
Bibliographic Details
Published inJournal of ambient intelligence and humanized computing Vol. 12; no. 11; pp. 10289 - 10303
Main Authors Shokeen, Jyoti, Rana, Chhavi
Format Journal Article
LanguageEnglish
Published Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2021
Springer Nature B.V
Subjects
Algorithms
Artificial Intelligence
Cold
Collaboration
Computational Intelligence
Engineering
Original Research
Ratings & rankings
Recommender systems
Robotics and Automation
Semantics
Social networks
Sparsity
Trust
User Interfaces and Human Computer Interaction
Matrix factorization
Social recommender system
Trust
Semantic
Online AccessGet full text

Cover

More Information
Summary:With the rapid advancement of Internet, e-commerce websites and social networks, people prefer to receive recommendations from their social friends rather than strangers. Also, the exponential evolution and use of online social networks has resulted in generation of enormous amount of information over web. The relationships between users in social networks are complex, vague and uncertain for computation. Adhering to the intuition that a user’s social network plays a prominent role in influencing the personal behavior of users on web, this paper proposes a trust and semantic-based social recommendation approach to remove cold-start issues. Social relationships are used to compute trust between users in the social networks. Trusted relations are used in addition to rating matrix to extract the implicit data. For each user, we also attempt to discover the top-k semantic friends because a user is connected to multiple friends on social networks who have different tastes. This proposed approach is superior to those traditional approaches that give equal weights to all users in social networks. One important advantage of this approach is consideration of social friends at different levels. Experimental results on real-world dataset demonstrate that our proposed approach outperforms some of the state-of-the-art recommendation approaches.
ISSN:1868-5137
1868-5145
DOI:10.1007/s12652-020-02806-1