WeChat Mini Program
Old Version Features

Hierarchical Query Classification in E-commerce Search

WWW 2024(2024)

Cited 1|Views44
Abstract
E-commerce platforms typically store and structure product information andsearch data in a hierarchy. Efficiently categorizing user search queries into asimilar hierarchical structure is paramount in enhancing user experience one-commerce platforms as well as news curation and academic research. Thesignificance of this task is amplified when dealing with sensitive querycategorization or critical information dissemination, where inaccuracies canlead to considerable negative impacts. The inherent complexity of hierarchicalquery classification is compounded by two primary challenges: (1) thepronounced class imbalance that skews towards dominant categories, and (2) theinherent brevity and ambiguity of search queries that hinder accurateclassification. To address these challenges, we introduce a novel framework that leverageshierarchical information through (i) enhanced representation learning thatutilizes the contrastive loss to discern fine-grained instance relationshipswithin the hierarchy, called ”instance hierarchy”, and (ii) a nuancedhierarchical classification loss that attends to the intrinsic label taxonomy,named ”label hierarchy”. Additionally, based on our observation that certainunlabeled queries share typographical similarities with labeled queries, wepropose a neighborhood-aware sampling technique to intelligently select theseunlabeled queries to boost the classification performance. Extensiveexperiments demonstrate that our proposed method is better thanstate-of-the-art (SOTA) on the proprietary Amazon dataset, and comparable toSOTA on the public datasets of Web of Science and RCV1-V2. These resultsunderscore the efficacy of our proposed solution, and pave the path toward thenext generation of hierarchy-aware query classification systems.
More
Translated text
Key words
Web Crawling
PDF
Bibtex
AI Read Science
Must-Reading Tree
Example
Generate MRT to find the research sequence of this paper
Data Disclaimer
The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn
Chat Paper
Summary is being generated by the instructions you defined