SessNet: A Hybrid Session-Based Recommender System

Omar Nada

doi:10.32920/17131607

Abstract

<div>Session-based recommendation is the task of predicting user actions during short online sessions. Previous work considers the user to be anonymous in this setting, with no past behavior history available. In reality, this is often not the case, and none of the existing approaches are flexible enough to seamlessly integrate user history when available. In this thesis, we propose a novel hybrid session-based recommender system to perform next-click prediction, which is able to take advantage of historical user preferences when accessible. Specifically, we propose SessNet, a deep profiling session-based recommender system, with a two-stage dichotomy. First, we use bidirectional transformers to model local and global session intent. Second, we concatenate any user information with the current session representation to feed to a feed-forward neural network to identify the next click. Historical user preferences are computed using the sequence-aware embeddings obtained from the first step, allowing us to better understand the users. We evaluate the efficacy of the proposed method using two benchmark datasets, YooChoose1/64 and Dignetica. Our experimental results show that SessNet outperforms state-of-the-art session-based recommenders on P@20 for both datasets.</div>

Highlights

In recent years, the social interactions have been growing vastly over the internet
After Recurrent Neural Network (RNN) has proven its strength in the domain of Session-based Recommender Systems (SRS), more research has been done to improve the existing models to increase accuracy and reduce running time
Our model is an extension to a standard RNN, which means that the model should be able to outperform a standard RNN

Summary

Introduction

The social interactions have been growing vastly over the internet. Social interactions can vary a lot in definition, where it could be considered mainly as two humans interacting with each other. With the presence of the World Wide Web, it has been much easier for people having interactions even if they don’t know each other These interactions can have different forms, one of them is interacting directly over social networks like Facebook or Twitter. This thesis will focus on the latter form of interactions where recommender system comes in place and plays a huge role in making these interactions as personalized as possible. Many popular websites such as Netflix, Amazon, Google, and Facebook rely a lot on the quality of their recommender systems to enhance the interactions on their websites. The amount of data has exploded giving more opportunities for new algorithms to play a role in personalizing services for the end users

Objectives

Methods

Results

Conclusion