Development of a cyberbullying detection model based on user personality (Big Five and Dark Triad), user sentiments and emotions using machine machine-learning algorithms. Tweets were used as the dataset in which they were classified as bully, spam, agressor or non-bully (i.e. normal).
PHyBR is a personalized book recommender that takes several users' characteristics into consideration, aiming to improve recommendation relevance. Characteristics include demographic details, personality traits, location, purchase intentions and sentiment of user reviews.
SentICal is an enhancement made to an existing sentiment calculator, specifically focusing on the manipulation of letter repetitions and elongated words.
The project improved mechanisms employed in feature extractions, tested against Amazon datasets.
The projects explored various mechanisms in which users' explicit and implicit data can be used to improve search results ranking.