Generating virtual ratings from chinese reviews to augment online recommendations

Abstract

Collaborative filtering (CF) recommenders based on User-Item rating matrix as explicitly obtained from end users have recently appeared promising in recommender systems. However, User-Item rating matrix is not always available or very sparse in some web applications, which has critical impact to the application of CF recommenders. In this article we aim to enhance the online recommender system by fusing virtual ratings as derived from user reviews. Specifically, taking into account of Chinese reviews' characteristics, we propose to fuse the self-supervised emotion-integrated sentiment classification results into CF recommenders, by which the User-Item Rating Matrix can be inferred by decomposing item reviews that users gave to the items. The main advantage of this approach is that it can extend CF recommenders to some web applications without user rating information. In the experiments, we have first identified the self-supervised sentiment classification's higher precision and recall by comparing it with traditional classification methods. Furthermore, the classification results, as behaving as virtual ratings, were incorporated into both user-based and item-based CF algorithms. We have also conducted an experiment to evaluate the proximity between the virtual and real ratings and clarified the effectiveness of the virtual ratings. The experimental results demonstrated the significant impact of virtual ratings on increasing system's recommendation accuracy in different data conditions (i.e., conditions with real ratings and without).

Recommender systems based on user reviews: the state of the art

“…2, we can infer that the reviewer has an overall positive opinion about this hotel. A simple way to estimate the overall opinion is to aggregate the sentiments of all of the opinion words that are contained in the review (Leung et al 2006;Zhang et al 2013). Alternatively, a machine learning algorithm (such as the naive Bayesian classifier or Support Vector Machine (SVM)) can be adopted to learn the opinion and classify it into a proper sentiment category (Pang et al 2002;Poirier et al 2010b).…”

“…Alternatively, a machine learning algorithm (such as the naive Bayesian classifier or Support Vector Machine (SVM)) can be adopted to learn the opinion and classify it into a proper sentiment category (Pang et al 2002;Poirier et al 2010b). The inferred overall opinions can then be converted into virtual ratings, which may take the role of real ratings in CF (Poirier et al 2010b;Zhang et al 2013) (see Sect. 4.2), or be used to enhance real ratings (Pero and Horváth 2013) (see Sect.…”

“…4.3.5). Emoticons (e.g., smiley and sad faces), as symbolic representations of emotions, can also be aggregated with opinion words to infer the reviewer's overall rating (Zhang et al 2013) (see Sect. 4.2).…”

“…In particular, information obtained from reviews is likely to benefit RS in the following three ways Garcia Esparza et al 2011;Hariri et al 2011;Jamroonsilp and Prompoon 2013;Levi et al 2012;McAuley and Leskovec 2013;Pero and Horváth 2013;Wang et al 2012;Yates et al 2008;Zhang et al 2013). …”

Recommendation Based on Contextual Opinions

“…Review-based recommenders mainly rely on advanced opinion mining techniques to infer the reviewers' overall opinion (called virtual rating [12]) or even multi-aspect ratings, which are then leveraged into the standard recommenders [3,5]. For instance, [8] developed a multi-label text classifier based on Support Vector Machine to reveal users' aspect-level evaluations of restaurants and generate recommendation through regression-based and clustering-based algorithms.…”