Explainable Outfit Recommendation with Joint Outfit Matching and Comment Generation

TL;DR

This paper introduces NOR, a neural network framework that jointly recommends outfits and generates comments to explain recommendations, leveraging visual features and multi-task learning.

Contribution

It presents a novel joint model for outfit matching and comment generation, integrating visual feature extraction with natural language explanation.

Findings

NOR outperforms state-of-the-art baselines in outfit recommendation accuracy.

Generated comments achieve high ROUGE and BLEU scores, comparable to human comments.

The joint training improves both recommendation quality and explanation relevance.

Abstract

Most previous work on outfit recommendation focuses on designing visual features to enhance recommendations. Existing work neglects user comments of fashion items, which have been proved to be effective in generating explanations along with better recommendation results. We propose a novel neural network framework, neural outfit recommendation (NOR), that simultaneously provides outfit recommendations and generates abstractive comments. NOR consists of two parts: outfit matching and comment generation. For outfit matching, we propose a convolutional neural network with a mutual attention mechanism to extract visual features. The visual features are then decoded into a rating score for the matching prediction. For abstractive comment generation, we propose a gated recurrent neural network with a cross-modality attention mechanism to transform visual features into a concise sentence. The two parts are jointly trained based on a multi-task learning framework in an end-to-end back-propagation paradigm. Extensive experiments conducted on an existing dataset and a collected real-world dataset show NOR achieves significant improvements over state-of-the-art baselines for outfit recommendation. Meanwhile, our generated comments achieve impressive ROUGE and BLEU scores in comparison to human-written comments. The generated comments can be regarded as explanations for the recommendation results. We release the dataset and code to facilitate future research.