Concept detection, a multifaceted term, involves identifying and extracting specific concepts within various data types, including text, images, and audio. This dissertation proposal is structured around three major objectives. The first objective focuses on document-level concept detection within patent documents, particularly in the field of Artificial Intelligence (AI). This objective aims to develop an automated patent landscaping system capable of identifying key AI concepts within vast patent datasets. Traditional patent landscaping is labor-intensive and expensive, making the need for automated landscape systems crucial. By leveraging neural models and advanced data representation techniques, this research aims to significantly improve the performance of existing systems. The second objective extends concept detection to the discourse level in educational environments. Specifically, it aims to detect granular concepts within teacher-student interactions in an intelligent immersive learning platform. The goal is to build a model that not only identifies the concepts being taught but also assesses the student's level of understanding. This objective integrates Automatic Speech Recognition (ASR) and natural language processing (NLP) techniques to analyze real-time educational conversations, potentially transforming teaching methodologies and curricular design. The third objective delves into concept discovery. In particular this problem focuses on identifying failure modes in learners, which indicate misunderstandings of concepts. This involves developing an educator-in-the-loop method to figure out when learners are failing to learn a concept. By analyzing argumentative features in student discourse, this research aims to cluster and identify patterns of misunderstanding, ultimately aiding in the development of personalized learning environment.

Tisa Islam Erana is a Ph.D. candidate at the Knight Foundation School of Computing and Information Sciences, Florida International University, under the supervision of Dr. Mark A. Finlayson. She is a part of the Cognition, Narrative, and Culture Laboratory (Cognac Lab) at FIU. Her research interest lies in Natural Language Processing, with a focus on concept detection. Tisa holds a bachelor’s degree in computer science from University of Dhaka, Bangladesh.