Semantic-based topic evaluation and application in information filtering
Tran Diem Hanh Nguyen · Queensland University of Technology · 2021
Topic modelling techniques are used to find the main themes in a collection of documents automatically. This thesis presents effective topic evaluation models to measure the quality of the discovered topics. The proposed techniques use human defined knowledge to solve the problems of evaluating topics in terms of semantic meaning of the topics. The thesis also proposed methods to modelling user interest based on the topic model generated from the user’s documents. The proposed techniques help to measure the quality of the topics and significantly improve the performance of text mining applications.