LSTM Networks for Online Cross-Network Recommendations

TL;DR

This paper introduces an online LSTM-based model with attention, higher-order interactions, and time-aware gates to enhance cross-network recommendations, addressing non-linear relationships and dynamic updates.

Contribution

It presents a novel multi-layered LSTM architecture tailored for online cross-network recommendation systems, incorporating mechanisms for long-term preferences, data sparsity, and temporal irregularities.

Findings

Outperforms state-of-the-art models in accuracy, diversity, and novelty.

Effectively captures user preference dynamics and data sparsity issues.

Demonstrates robustness across Twitter, Google Plus, and YouTube data.

Abstract

Cross-network recommender systems use auxiliary information from multiple source networks to create holistic user profiles and improve recommendations in a target network. However, we find two major limitations in existing cross-network solutions that reduce overall recommender performance. Existing models (1) fail to capture complex non-linear relationships in user interactions, and (2) are designed for offline settings hence, not updated online with incoming interactions to capture the dynamics in the recommender environment. We propose a novel multi-layered Long Short-Term Memory (LSTM) network based online solution to mitigate these issues. The proposed model contains three main extensions to the standard LSTM: First, an attention gated mechanism to capture long-term user preference changes. Second, a higher order interaction layer to alleviate data sparsity. Third, time aware LSTM cell gates to capture irregular time intervals between user interactions. We illustrate our solution using auxiliary information from Twitter and Google Plus to improve recommendations on YouTube. Extensive experiments show that the proposed model consistently outperforms state-of-the-art in terms of accuracy, diversity and novelty.