My research focuses on building user knowledge models for adaptive and personalized recommendation. User models rely on direct or indirect evidence from users like quiz performance in tutoring systems, item clicks in recommendation systems and like reaction on social media sites. However, there are scenarios when these feedback are missing or are not enough. This is where I am digging in and trying to use user behaviours likes reading behavior, user reviews and purpose of sharing a post to build user knowledge models. In my recent work, I tried to improve student models by inferring better student knowledge state through student reading behaviour. Currently I am working on utilizing ovarian cancer forum users' post behaviour to build their within-domain knowledge model.I love exploring new possibilities in Natural Language Processing and Information Retrieval. In my recent exploration I implemented a faceted abstract summarization system. Some of my work includes Keyphrase extraction, Keyphrase representation and infering Domain Specificity.website: https://kthaker.com