Contextual Sequence Modeling for Recommendation with Recurrent Neural Networks

Abstract

Recommendations can greatly benefit from good representations of the user state at recommendation time. Recent approaches that leverage Recurrent Neural Networks (RNNs) for session-based recommendations have shown that Deep Learning models can provide useful user representations for recommendation. However, current RNN modeling approaches summarize the user state by only taking into account the sequence of items that the user has interacted with in the past, without taking into account other essential types of context information such as the associated types of user-item interactions, the time gaps between events and the time of day for each interaction. To address this, we propose a new class of Contextual Recurrent Neural Networks for Recommendation (CRNNs) that can take into account the contextual information both in the input and output layers and modifying the behavior of the RNN by combining the context embedding with the item embedding and more explicitly, in the model dynamics, by parametrizing the hidden unit transitions as a function of context information. We compare our CRNNs approach with RNNs and non-sequential baselines and show good improvements on the next event prediction task.

A Simple Convolutional Generative Network for Next Item Recommendation

“…In this section we detail our experiments, report results for several data sets, and compare our model (called NextItNet) with the wellknown RNN-based model GRURec [15,28] and the state-of-the-art CNN-based model Caser. Note that (1) since the main contributions in this paper do not focus on combining various features, we omit the comparison with content-or context-based sequential recommendation models, such as the 3D CNN recommender [30] and other RNN variants [9,20,25,27]; (2) the GRURec baseline could be regarded as the state-of-the-art Improved GRURec [28] when dealing with the long-range session data sets because our main data augmentation technique for the two baseline models follows the same way in Improved GRURec.…”

“…For example, a Gated Recurrent Unit (GRURec) architecture with a ranking loss was proposed by [15] for session-based recommendation. In the follow-up papers, various RNN variants have been designed to extend the typical one for different application scenarios, such as by adding personalization [25], content [9] and contextual features [27], attention mechanism [7,20] and different ranking loss functions [14].…”

Siamese Cookie Embedding Networks for Cross-Device User Matching

“…For the SCEmNet implementation, we fixed the different filter sizes to be (2,3,4,5,6,10) and the number of filters per size at 40. e dropout probability was set at 0.5 for training and 0 for test. Negative pairs were sampled from the set of candidate pairs not present in the list of ground truth pairs.…”

Contextual Hybrid Session-Based News Recommendation With Recurrent Neural Networks

“…Contextual information was used in combination with RNNs, for example, in [70] or [71]. In [70], the authors consider not only the sequence of events when making predictions but also the type of the event, the time gaps between events, or the time of the day of an event, leading to what they call Contextual Recurrent Neural Networks for Recommendation (CRNN). Similarly, Twardowski [71] considers time as a contextual factor that is combined with item information within a hybrid approach.…”