A framework of a recommendation system utilizing expert groups on a social network

Abstract

A social network is used as a mechanism to link people together to solicit and relay recommendations from one another. However, in a large social network where most people would have hundreds of acquaintances and millions of people within the social network, relying solely on the recommendations obtained through a search that involves a significant number of people within a network, which may not be the most practical and economical option. A solution to this is to limit the number of people between two people within a social network, between the person soliciting a recommendation and a person potentially providing a recommendation. To compensate for the potential loss of recommendations as a result of the limit, a mechanism to compliment the recommendation system, known as expert groups, is created. Expert groups are a collection of people with a common expertise in a common knowledge area and a certain degree of like-mindedness. People within these expert groups can provide recommendations on issues within the common knowledge area. The proposed framework uses software agents to model the behavior of people when soliciting recommendations and providing recommendations. © 2011 Springer-Verlag.