Learning Recommender Systems with Soft Target: A Decoupled Perspective

TL;DR

This paper introduces a decoupled soft label optimization framework for recommender systems, improving the handling of positive and negative feedback by modeling latent interests and adjusting importance, validated through extensive experiments.

Contribution

It proposes a novel decoupled soft label framework and loss function for better feedback modeling in recommender systems, along with a soft-label generation algorithm based on label propagation.

Findings

Enhanced recommendation accuracy across multiple models

Effective modeling of latent user interests

Demonstrated generality on public datasets

Abstract

Learning recommender systems with multi-class optimization objective is a prevalent setting in recommendation. However, as observed user feedback often accounts for a tiny fraction of the entire item pool, the standard Softmax loss tends to ignore the difference between potential positive feedback and truly negative feedback. To address this challenge, we propose a novel decoupled soft label optimization framework to consider the objectives as two aspects by leveraging soft labels, including target confidence and the latent interest distribution of non-target items. Futhermore, based on our carefully theoretical analysis, we design a decoupled loss function to flexibly adjust the importance of these two aspects. To maximize the performance of the proposed method, we additionally present a sensible soft-label generation algorithm that models a label propagation algorithm to explore users' latent interests in unobserved feedback via neighbors. We conduct extensive experiments on various recommendation system models and public datasets, the results demonstrate the effectiveness and generality of the proposed method.