A rating aggregation method for generating product reputations

Abstract

Many websites offer the opportunity for customers to rate items and then use customers' ratings to generate items reputation, which can be used later by other users for decision making purposes. The aggregated value of the ratings per item represents the reputation of this item. The accuracy of the reputation scores is important as it is used to rank items. Most of the aggregation methods didn't consider the frequency of distinct ratings and they didn't test how accurate their reputation scores over different datasets with different sparsity. In this work we propose a new aggregation method which can be described as a weighted average, where weights are generated using the normal distribution. The evaluation result shows that the proposed method outperforms stateof-the-art methods over different sparsity datasets.

A Reputation-Enhanced Recommender System

“…The weight can represent the user's reputation score, the time when the rating was given, or the distance between the current reputation score and the rating received [6,7]. Abdel-Hafez et al [20,21] used the normal-distribution to generate weighted average reputation model which explicitly reflects the distribution of ratings of items.…”

Item Reputation-Aware Recommender Systems

“…The proposed method uses two ranked lists of items; the first list is generated by a recommender system, such as the user-based CF recommender system [6], and the second list is generated based on item reputations calculated using a reputation model, such as the NDRU reputation model [7]. The two ranked lists are then combined to enhance the accuracy of the recommendations.…”

“…The recommender used with the three merging methods was the userbased CF [6] and the reputation model was the NDRU reputation model which was calculated using equation (3) [7]. We chose this model due to its stable and satisfactory performance regardless of the sparsity of the dataset.…”

“…The generality is one of the advantages of our proposed method, as any recommendation or reputation method can be used in conjunction. We employed a user-based CF method [6] and the Normal Distribution based Reputation with Uncertainty (NDRU) model [7]. We also tested other recommendation methods such as BiasedMF [8], SVD++ [8], and RegSVD [9].…”

Utilizing Voting Systems for Ranking User Tweets

“…1-Movie reputation (MR): we calculate the reputation of a movie based-on the IMDB users' ratings for this movie using a normal distribution based reputation model with uncertainty [19]. This model is described as a weighted average method where the weights reflect the distribution of ratings in the overall score.…”