Master Thesis Student
Thesis topic: SIMT: A Semantic Interest Modeling Toolkit
Related project: RIMA
Thesis duration: 02/2020 - 08/2020
Interest modeling refers to inferring topics of interests for users based on the information they generated. By capturing user's interests, personalized services can be achieved. For example, people can find like-minded friends with same interests, researchers with similar interests can collaborate to work on their publications, advertisers can locate the potential customers precisely. Thus, it is an important step for content-based recommendation services. Text mining techniques such as keyphrase extraction can support the generation of the user's interest model based on the textual user-generated content. However, due to the lack of semantic knowledge, it is always associated with semantic problems such as overgeneration or redundancy errors, etc. Besides, the similarity of interest models is needed when we perform the recommendation service. However, the traditional similarity measures failed to capture the semantic relatedness between the interest models and lead to inaccurate recommendation results.
The aim of this thesis is to develop a Python-based interest modeling toolkit to address the semantic issues and to provide a React application to effectively generate interest models and compute their similarities based on the toolkit. Moreover, the effectiveness evaluation should be conducted to validate the proposed semantic interest modeling approach.