General Item Representation Learning for Cold-start Content Recommendations

TL;DR

This paper introduces a domain-agnostic, Transformer-based item representation learning framework that effectively addresses cold-start recommendation challenges by leveraging multimodal feature alignment without relying on classification labels.

Contribution

It proposes a novel, end-to-end trainable framework that utilizes multimodal alignment for cold-start recommendations, surpassing state-of-the-art baselines across multiple domains.

Findings

Better preserves fine-grained user preferences.

Universally applicable to multiple large-scale domains.

Outperforms existing state-of-the-art methods.

Abstract

Cold-start item recommendation is a long-standing challenge in recommendation systems. A common remedy is to use a content-based approach, but rich information from raw contents in various forms has not been fully utilized. In this paper, we propose a domain/data-agnostic item representation learning framework for cold-start recommendations, naturally equipped with multimodal alignment among various features by adopting a Transformer-based architecture. Our proposed model is end-to-end trainable completely free from classification labels, not just costly to collect but suboptimal for recommendation-purpose representation learning. From extensive experiments on real-world movie and news recommendation benchmarks, we verify that our approach better preserves fine-grained user taste than state-of-the-art baselines, universally applicable to multiple domains at large scale.