New Item Consumption Prediction Using Deep Learning

TL;DR

This paper introduces PISA, a deep learning-based content approach for predicting purchase intent in cold start scenarios, outperforming existing methods especially with new items and imbalanced data.

Contribution

The paper presents PISA, a novel deep learning algorithm that improves cold start purchase intent prediction and handles data imbalance effectively.

Findings

PISA outperforms baseline in new item scenarios.

PISA handles highly imbalanced datasets well.

Combining PISA with baseline improves overall performance.

Abstract

Recommendation systems have become ubiquitous in today's online world and are an integral part of practically every e-commerce platform. While traditional recommender systems use customer history, this approach is not feasible in 'cold start' scenarios. Such scenarios include the need to produce recommendations for new or unregistered users and the introduction of new items. In this study, we present the Purchase Intent Session-bAsed (PISA) algorithm, a content-based algorithm for predicting the purchase intent for cold start session-based scenarios. Our approach employs deep learning techniques both for modeling the content and purchase intent prediction. Our experiments show that PISA outperforms a well-known deep learning baseline when new items are introduced. In addition, while content-based approaches often fail to perform well in highly imbalanced datasets, our approach successfully handles such cases. Finally, our experiments show that combining PISA with the baseline in non-cold start scenarios further improves performance.