Review Polarity-wise Recommender
This work addresses a specific issue in recommender systems for e-commerce or review platforms by improving accuracy through polarity-aware modeling, representing an incremental advance over existing methods.
The paper tackles the problem that existing review-involved recommender systems ignore opposite aspect information in positive and negative reviews, which can mislead user preference modeling, by proposing a Review Polarity-wise Recommender (RPR) model that separately processes reviews by polarity and uses aspect-aware weighting to address imbalance, achieving superior performance on eight benchmark datasets compared to state-of-the-art baselines.
Utilizing review information to enhance recommendation, the de facto review-involved recommender systems, have received increasing interests over the past few years. Thereinto, one advanced branch is to extract salient aspects from textual reviews (i.e., the item attributes that users express) and combine them with the matrix factorization technique. However, existing approaches all ignore the fact that semantically different reviews often include opposite aspect information. In particular, positive reviews usually express aspects that users prefer, while negative ones describe aspects that users reject. As a result, it may mislead the recommender systems into making incorrect decisions pertaining to user preference modeling. Towards this end, in this paper, we propose a Review Polarity-wise Recommender model, dubbed as RPR, to discriminately treat reviews with different polarities. To be specific, in this model, positive and negative reviews are separately gathered and utilized to model the user-preferred and user-rejected aspects, respectively. Besides, in order to overcome the imbalance problem of semantically different reviews, we also develop an aspect-aware importance weighting approach to align the aspect importance for these two kinds of reviews. Extensive experiments conducted on eight benchmark datasets have demonstrated the superiority of our model as compared to a series of state-of-the-art review-involved baselines. Moreover, our method can provide certain explanations to the real-world rating prediction scenarios.