Multi-Task Feature Learning for Knowledge Graph Enhanced Recommendation

TLDR

Collaborative filtering suffers from sparsity and cold start, so side information such as knowledge graphs is often used to improve recommender systems. The authors introduce MKR, a multi‑task feature learning framework that enhances recommendation with knowledge graph side information. MKR is a deep end‑to‑end model that jointly learns knowledge‑graph embeddings and recommendation, using cross‑compress units to share latent features and capture high‑order interactions between items and entities. Experiments on movie, book, music, and news datasets show that MKR outperforms state‑of‑the‑art baselines and remains robust under sparse user‑item interactions.

Abstract

Collaborative filtering often suffers from sparsity and cold start problems in real recommendation scenarios, therefore, researchers and engineers usually use side information to address the issues and improve the performance of recommender systems. In this paper, we consider knowledge graphs as the source of side information. We propose MKR, a Multi-task feature learning approach for Knowledge graph enhanced Recommendation. MKR is a deep end-to-end framework that utilizes knowledge graph embedding task to assist recommendation task. The two tasks are associated by crosscompress units, which automatically share latent features and learn high-order interactions between items in recommender systems and entities in the knowledge graph. We prove that crosscompress units have sufficient capability of polynomial approximation, and show that MKR is a generalized framework over several representative methods of recommender systems and multi-task learning. Through extensive experiments on real-world datasets, we demonstrate that MKR achieves substantial gains in movie, book, music, and news recommendation, over state-of-the-art baselines. MKR is also shown to be able to maintain satisfactory performance even if user-item interactions are sparse.