Search Papers | Poster Sessions | All Posters
Poster A93 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
Towards Semantic Classification of Dialog using Contextual Prediction Networks
Alex Doboli1, Gnaneswar Villuri1; 1Stony Brook University
Semantic classification distinguishes inputs based on their meaning (e.g., interpretation) not their static features, as in traditional classification. Existing transformer models seem to have limited capabilities for semantic classification. This paper presents our ongoing work on the semantic classification of the dialog sentences produced by human subjects during problem solving, including the used data set, and the gained insight from using transformer models for classification. A new theoretical model, called Contextual Prediction Networks, is suggested for semantic classification of dialog sentences.
Keywords: semantic classification dialog explainability transformer models