A Comparative Study of Multi-Domain Intent Classification Models Using Explainable Artificial Intelligence
Disciplines
Artificial Intelligence and Robotics | Computer Sciences
Abstract (300 words maximum)
Intent classification is a critical task in enabling the ability of intelligent systems to identify the purpose of text generated by users. While existing research has achieved strong results using transformer-based models such as BERT, most studies only focus on a single domain such as emails, text messages, or social media posts, this limits the generalization and interpretability of intent recognition systems across diverse communication settings. This research presents a comparative analysis of traditional machine learning and transformer-based models for multi domain intent classification, integrating explainable artificial intelligence techniques to enhance model transparency. This study will explore applications such as identifying potentially harmful or criminal intent in online forums, which could prevent future harm and distinguish genuine purchase intent from casual search in e-commerce site visits, which will improve targeted recommendations. By combining a multi-domain evaluation with explainable analysis, this work contributes to a framework that can be used for all domains that develops accurate and interpretable intent.
Use of AI Disclaimer
no
Academic department under which the project should be listed
CCSE – Computer Science
Primary Investigator (PI) Name
Md Abdullah Al Hafiz Khan
A Comparative Study of Multi-Domain Intent Classification Models Using Explainable Artificial Intelligence
Intent classification is a critical task in enabling the ability of intelligent systems to identify the purpose of text generated by users. While existing research has achieved strong results using transformer-based models such as BERT, most studies only focus on a single domain such as emails, text messages, or social media posts, this limits the generalization and interpretability of intent recognition systems across diverse communication settings. This research presents a comparative analysis of traditional machine learning and transformer-based models for multi domain intent classification, integrating explainable artificial intelligence techniques to enhance model transparency. This study will explore applications such as identifying potentially harmful or criminal intent in online forums, which could prevent future harm and distinguish genuine purchase intent from casual search in e-commerce site visits, which will improve targeted recommendations. By combining a multi-domain evaluation with explainable analysis, this work contributes to a framework that can be used for all domains that develops accurate and interpretable intent.