Supercharging recommender systems using taxonomies for learning user purchase behavior

Abstract

Recommender systems based on latent factor models have been effectively used for understanding user interests and predicting future actions. Such models work by projecting the users and items into a smaller dimensional space, thereby clustering similar users and items together and subsequently compute similarity between unknown user-item pairs. When user-item interactions are sparse (sparsity problem) or when new items continuously appear (cold start problem), these models perform poorly. In this paper, we exploit the combination of taxonomies and latent factor models to mitigate these issues and improve recommendation accuracy. We observe that taxonomies provide structure similar to that of a latent factor model: namely, it imposes human-labeled categories (clusters) over items. This leads to our proposed taxonomy-aware latent factor model (TF) which combines taxonomies and latent factors using additive models. We develop efficient algorithms to train the TF models, which scales to large number of users/items and develop scalable inference/recommendation algorithms by exploiting the structure of the taxonomy. In addition, we extend the TF model to account for the temporal dynamics of user interests using highorder Markov chains. To deal with large-scale data, we develop a parallel multi-core implementation of our TF model. We empirically evaluate the TF model for the task of predicting user purchases using a real-world shopping dataset spanning more than a million users and products. Our experiments demonstrate the benefits of using our TF models over existing approaches, in terms of both prediction accuracy and running time.We resolve the aforementioned challenges by using a taxonomy over items which is available for many consumer products, e.g., PriceGrabber shopping taxonomy [3], music taxonomy [16], movie taxonomy (genre, director and so on). A fragment of the Yahoo! shopping taxonomy (that we use in our paper) is shown in Figure 1. Since taxonomies are designed by humans, they capture knowledge about the domain at hand, independent of the training data [13] and hence provide scope for improving prediction. As such, taxonomies can provide lineage for items in terms of categories and their ancestors in the taxonomy, thus they help to ameliorate the sparsity issues that are common with online shopping data. Also, taxonomies can help in dealing with the cold start problem because even though the set of individual products/items is highly dynamic,

Inferring continuous dynamic social influence and personal preference for temporal behavior prediction

“…In the literature, it gets extensively studied in the area of recommender systems [1,16], which focuses on recommending the most appropriate items to users based on their past adoption behavior data. The collaborative filtering technique [20] has been widely adopted and numerous methods have been proposed [10,26,18,13,4]. With the emergence of online social networks, the social relationships are found beneficial for such tasks [11,9,14,27], because friends may influence each other and thus tend to exhibit similar behaviors [7].…”

Exploiting User Preference for Online Learning in Web Content Optimization Systems

“…Furthermore, in the similar spirit as in the pairwise learning-torank algorithm for search, using pairwise preferences for devising personalized recom-mendation algorithm also has been used for recommender systems, e.g., [Rendle et al 2009;Takács and Tikk 2012;Yang et al 2011;Kanagal et al 2012]. To address the problem of sparse learning samples in recommender system, some previous studies introduced factorized parametrized models which can make estimation with a small sample size.…”

“…For example, [Rendle et al 2010] proposed a Markov Chain model which factorizes the transition matrix such that transitions can be estimated with little or even no observations. In addition, [Koren 2009;Xiong et al 2010;Kanagal et al 2012] have considered incorporating some temporal variations of user preferences in the recommender system.…”

RECATHON: A Middleware for Context-Aware Recommendation in Database Systems

“…Recently, recommender systems have grabbed researchers' attention in both industry [2], [7], [9], [19] and academia [14], [13], [24], [28], [29]. The main goal of a recommender system is to suggest new and interesting items to users from a large pool of items.…”