Inter-Session Modeling for Session-Based Recommendation

Abstract

In recent years, research has been done on applying Recurrent Neural Networks (RNNs) as recommender systems. Results have been promising, especially in the session-based setting where RNNs have been shown to outperform state-of-the-art models. In many of these experiments, the RNN could potentially improve the recommendations by utilizing information about the user's past sessions, in addition to its own interactions in the current session. A problem for session-based recommendation, is how to produce accurate recommendations at the start of a session, before the system has learned much about the user's current interests. We propose a novel approach that extends a RNN recommender to be able to process the user's recent sessions, in order to improve recommendations. This is done by using a second RNN to learn from recent sessions, and predict the user's interest in the current session. By feeding this information to the original RNN, it is able to improve its recommendations. Our experiments on two different datasets show that the proposed approach can significantly improve recommendations throughout the sessions, compared to a single RNN working only on the current session. The proposed model especially improves recommendations at the start of sessions, and is therefore able to deal with the cold start problem within sessions.

Deep Learning-Based Sequential Recommender Systems: Concepts, Algorithms, and Evaluations

“…In the next sections, we will further summarize and evaluate the factors that might impact the performance of a DL-based model, which is expected to better guide future research. [41], [42] TB RNN [8], [43], [44], [51]- [56] [13], [45]- [50], [57], [58] CNN [24], [25], [60], [61] MLP [69], [71] att.…”

“…[57] used RNN for collaborative filtering and considered two different objective functions in the RNN model: categorical cross-entropy (CCE) and Hinge, where CCE is the most widely used in language modeling, and Hinge is extended from the objective function of SVMs. [58] uses multi-layer GRU network to capture sequential dependencies and user interest from both inter-session and intra-session level.…”

Time is of the Essence

“…Furthermore, the time modeling is on the inter-basket/session time-gaps, and not the time between each selection. It means using a single level RNN for inter-session modeling directly, in contrast to [19] and [16], both implementing some form of inter-session modeling in one of the two levels of their hierarchical RNN models. The final model's recommendation capabilities was compared with two factorization based baselines and one based on neural networks.…”

“…The evaluation is performed on two different datasets as in [19]: the LastFM dataset [1] containing listening habits of users on the music website Last.fm and the Reddit dataset 2 , on user activity on the social news aggregation and discussion website Reddit. Last.fm is a music website where users can keep track of the songs they listen to as well as sharing this with their peers.…”

“…The data were first preprocessed by removing noisy and irrelevant data and defining markers in the data following the steps in [19]. The Reddit dataset contains a log of user interaction on different subreddits (sub-forums), with timestamps.…”