‎Provenance Based Trust Boosted Recommender System Using Boosted Vector Similarity Measure

‎As users in an online social network are overwhelmed by the abundant amount of information‎, ‎it is very hard to retrieve the preferred or required content‎. ‎In this context‎, ‎an online recommender system helps to filter and recommend content such as people,items or services‎. ‎But‎, ‎in a real scenario‎, ‎people rely more on recommendations‎ ‎from trusted sources than distrusting sources‎. ‎Though‎, ‎there are many trust based recommender systems that exist‎, ‎it lag in prediction error‎. ‎In order to improve the accuracy of the prediction‎, ‎this paper proposes a Trust-Boosted Recommender System (TBRS)‎. ‎Since‎, ‎the provenance derives the trust in a better way than other approaches‎, ‎TBRS is built‎ ‎from the provenance concept‎. ‎The proposed recommender system takes the provenance based fuzzy rules which were derived from the Fuzzy Decision Tree‎. ‎TBRS then computes the multi-attribute vector similarity score and boosts the score with trust weight‎. ‎This system is tested on the book-review dataset to recommend the top-k trustworthy reviewers.The performance of the proposed method is evaluated in terms of MAE and RMSE‎. ‎The result shows that the error value of boosted similarity is lesser than without boost‎. ‎The reduced error rates of the Jaccard‎, ‎Dice and Cosine similarity measures are 18\%‎, ‎15\% and 7\% respectively‎. ‎Also‎, ‎when the model is subjected to failure analysis‎, ‎it gives better performance for unskewed data than slewed data‎. ‎The models fbest‎, ‎average and worst case predictions are 90\%‎, ‎50\% and $<$23\% respectively‎.