Leveraging User Behaviour to Enhance LLMs for Query Analysis in E-commerce
Leveraging User Behaviour to Enhance LLMs for Query Analysis in E-commerce
Abstract
The problem with using LLMs (Large Language Models) for query analysis in e-commerce search is that the accuracy rate can at most reach up to 90%, due to the lack of domain specific information. However, based on mining query tags from user orders, although the accuracy can reach over 95%, the coverage rate of query tags is limited by the amount of user behavior logs, especially when the business scope is small. In this article, we propose a method for constructing training dataset for query tags, which combines the data by LLMs with the data based on user order logs. Models trained on this dataset can achieve a coverage rate of over 95% and an accuracy rate of over 95% for query tagging.
1. Introduction
Query analysis is the task of query tagging, such as query synonyms, query categories, etc.
2. Method
2.1 Prompt LLMs for Query Analysis
2.2 User Behaviour Based Query Analysis
3. Evaluation
4. Discussion
4.1 Motivation
5. Conclusion
Reference
\bibitem{ref1}
\bibitem{ref2}
\bibitem{ref3}
\bibitem{ref4}
\bibitem{ref5}