Effective question-intent understanding plays an important role in enhancing the performance of Question-Answering (QA) and Search systems. Previous research in open-domain QA has highlighted the value of intent taxonomies in comprehending data and facilitating answer generation and evaluation. However, existing taxonomies have limitations for specific domains. We’re interested in question intent for e-commerce scenarios where questions are specific to shopping activities.
To address such limitations, we propose the adoption of a bespoke strategy for the e-commerce domain. We introduce an E-commerce Question Answering (EQA) taxonomy designed to encapsulate the unique aspects of e-commerce queries. Our empirical analyses validate the EQA taxonomy’s ability to more accurately represent users’ information needs in shopping scenarios. Further, we employed instruction fine-tuning to develop an intent classifier capable of categorizing questions following EQA taxonomy. Our result shows that EQA can provide clear guidance for intent classification for e-commerce queries. Finally, our approach shows that it is possible to build a domain-specific taxonomy and associated classifiers that can be used in different applications.
Research areas
