No abstract

This paper leverages heterogeneous auxiliary information to address the data sparsity problem of recommender systems. We propose a model that learns a shared feature space from heterogeneous data, such as item descriptions, product tags and online purchase history, to obtain better predictions. Our model consists of autoencoders, not only for numerical and categorical data, but also for sequential data, which enables capturing user tastes, item characteristics and the recent dynamics of user preference. We learn the autoencoder architecture for each data source independently in order to better model their statistical properties. Our evaluation on two MovieLens datasets and an ecommerce dataset shows that mean average precision and recall improve over state-of-the-art methods.

Rakuten Ichiba uses a taxonomy to organize the items it sells. Currently, the taxonomy classes that are relevant in terms of profit generation and difficulty of exploration are being manually extended with data properties deemed helpful to create pages that improve the user search experience and ultimately the conversion rate. In this paper we present a scalable approach that aims to automate this process, automatically selecting the relevant and semantically homogenous subtrees in the taxonomy, extracting from semi-structured text in items descriptions a core set of properties and a popular subset of their ranges, then extending the covered range using relational similarities in free text. Additionally, our process automatically tags the items with the new semantic information and exposes them as RDF triples. We present a set of experiments showing the effectiveness of our approach in this business context.

Existing algorithms aiming to learn a binary classifier from positive (P) and unlabeled (U) data require estimating the class prior or label noise ahead of building a classification model. However, the estimation and classifier learning are normally conducted in a pipeline instead of being jointly optimized. In this paper, we propose to alternatively train the two steps using reinforcement learning. Our proposal adopts a policy network to adaptively make assumptions on the labels of unlabeled data, while a classifier is built upon the output of the policy network and provides rewards to learn a better policy. The dynamic and interactive training between the policy maker and the classifier can exploit the unlabeled data in a more effective manner and yield a significant improvement in terms of classification performance. Furthermore, we present two different approaches to represent the actions taken by the policy. The first approach considers continuous actions as soft labels, while the other uses discrete actions as hard assignment of labels for unlabeled examples. We validate the effectiveness of the proposed method on two public benchmark datasets as well as one ecommerce dataset. The results show that the proposed method is able to consistently outperform state-of-the-art methods in various settings.

No abstract