Effect of influential users on recommendation

Relevant information stored in boundless pool of data source are required for the recommendation provided for users in recommender systems. Current recommender systems still suffer from inaccurate or erroneous predictions for users. This may be due to lack of consensus between users who provide different opinions on items after purchase. However, it is possible that this problem might be due to the users having no/few knowledge on the items or they might have had diverse reasons for previous purchase of the items. Therefore, they decide to either provide untruthful opinions on the items or to not even provide their opinions on the items. This demo paper presents a proposed approach to recommendation, where trust information from the social network can be used to motivate or influence users to contribute their opinions for future recommendation. A new trust metric based on trust features such as familiarity and experience value will be used to identify influential users who will control information flow and motivate the members in their community.