A Semi-supervised Scalable Unified Framework for E-commerce Query Classification

TL;DR

This paper introduces a semi-supervised, scalable, unified framework for e-commerce query classification that leverages knowledge, label semantics, and structure enhancements to improve accuracy and efficiency across multiple subtasks.

Contribution

The paper proposes a novel SSUF framework that unifies multiple query classification subtasks with modular enhancements, addressing data scarcity and efficiency issues in e-commerce applications.

Findings

Significantly outperforms state-of-the-art models in offline tests.

Achieves notable improvements in online A/B experiments.

Effectively handles short, context-limited queries with knowledge and label enhancements.

Abstract

Query classification, including multiple subtasks such as intent and category prediction, is vital to e-commerce applications. E-commerce queries are usually short and lack context, and the information between labels cannot be used, resulting in insufficient prior information for modeling. Most existing industrial query classification methods rely on users' posterior click behavior to construct training samples, resulting in a Matthew vicious cycle. Furthermore, the subtasks of query classification lack a unified framework, leading to low efficiency for algorithm optimization. In this paper, we propose a novel Semi-supervised Scalable Unified Framework (SSUF), containing multiple enhanced modules to unify the query classification tasks. The knowledge-enhanced module uses world knowledge to enhance query representations and solve the problem of insufficient query information. The label-enhanced module uses label semantics and semi-supervised signals to reduce the dependence on posterior labels. The structure-enhanced module enhances the label representation based on the complex label relations. Each module is highly pluggable, and input features can be added or removed as needed according to each subtask. We conduct extensive offline and online A/B experiments, and the results show that SSUF significantly outperforms the state-of-the-art models.