Knowledge Perceived Multi-modal Pretraining in E-commerce

TL;DR

This paper introduces K3M, a multi-modal pretraining method for E-commerce product data that incorporates knowledge modality to improve robustness against missing or noisy modalities, leading to better performance.

Contribution

K3M is a novel multi-modal pretraining approach that integrates knowledge modality and models modality interactions to enhance robustness in E-commerce scenarios.

Findings

K3M outperforms baseline methods under modality-noise conditions.

K3M achieves significant improvements on real-world E-commerce datasets.

The method effectively handles missing and noisy modalities in product data.

Abstract

In this paper, we address multi-modal pretraining of product data in the field of E-commerce. Current multi-modal pretraining methods proposed for image and text modalities lack robustness in the face of modality-missing and modality-noise, which are two pervasive problems of multi-modal product data in real E-commerce scenarios. To this end, we propose a novel method, K3M, which introduces knowledge modality in multi-modal pretraining to correct the noise and supplement the missing of image and text modalities. The modal-encoding layer extracts the features of each modality. The modal-interaction layer is capable of effectively modeling the interaction of multiple modalities, where an initial-interactive feature fusion model is designed to maintain the independence of image modality and text modality, and a structure aggregation module is designed to fuse the information of image, text, and knowledge modalities. We pretrain K3M with three pretraining tasks, including masked object modeling (MOM), masked language modeling (MLM), and link prediction modeling (LPM). Experimental results on a real-world E-commerce dataset and a series of product-based downstream tasks demonstrate that K3M achieves significant improvements in performances than the baseline and state-of-the-art methods when modality-noise or modality-missing exists.