Title: Ant-based friends recommendation in social tagging systems

Authors: Punam Bedi; Ravish Sharma

Addresses: Department of Computer Science, Faculty of Mathematical Sciences, Opposite Daulat Ram College, University of Delhi, Delhi-110007, India ' Department of Computer Science, PGDAV College, University of Delhi, Nehru Nagar, Delhi-110065, India

Abstract: Aggregated tagging behaviour in social tagging systems provides valuable information about the interests of the users which may help in broadening their perspective and expand their network of friends. This paper proposes an ant-based friends recommendation (AbFR) approach to recommend friends to the target user. In AbFR, prominent-N tags are recommended to the target user for a specific resource using tag co-occurrence graph where weights on the edges are computed based on pheromone updating strategy known from ant algorithms. Each target user's trust graph containing friends is maintained based on implicit feedback over a period of time. Optimal trust path is computed from the combined trust graph of all the users and friends are recommended to the target user based on foraging behaviour of biological ants. Experimental study conducted with delicious.com, a social bookmarking website shows significant improvement in discovering new friends.

Keywords: social tagging systems; ant colony metaphor; pheromone updating; co-occurrence graph; delicious.com; friends recommendation; user interests; pheromone updating strategy; trust path; foraging behaviour; recommender systems; social bookmarking websites.

DOI: 10.1504/IJSI.2015.072888

International Journal of Swarm Intelligence, 2015 Vol.1 No.4, pp.321 - 343

Accepted: 18 Dec 2013
Published online: 06 Nov 2015 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article