A method to support dynamic domain model based on user interests for effective language learning

摘要

This study aims to explore a method that can generate a dynamic domain model based on the user's interests and status updates. To get the most relevant interests of an individual the following algorithms were used after the study by M. Timonen: Inverse Fragment Length, Category Probability, Binormal Separation, Fragment Length Weighted Category Distribution and Time Sensitive Term Weighting. This study has shown that it is possible to obtain a dynamic user model representation through their social media profile. This was done by implementing a proof-of-concept application on news recommender system. Future work for this study includes evaluating this method in language learning.