Natural Language Processing, Explainable AI, User Modeling and Adaptation, and Intelligent User Interfaces.
Decision Support System that does primary patient assessment, UofA.
The goal is for an intelligent agent to listen to a patient's concern and tell them what should be their next step. (e.g. go to a doctor, come to the emergency, stay home and observe , etc.)
Detecting Propaganda from News Articles, UofA.
Works on solving the first problem of SemEval Task-11, detecting propaganda phrases from news articles.
Experiments on various RNN and BERT-based models (BILSTM-CRF, BERT-CRF etc.) to solve the problem.
Models beat the performance from the original paper by Martino et. al.
Automatic Removal of Outliers to Improve Density-Based Clustering Performance, UofA.
The goal is to automatically remove outliers from data while maintaining inliers as much as possible without any prior knowledge.