Term Extraction for Micro-domain Ontology in Personalized Search


To improve the search quality of a search engine so that it can better understand and return the information to meet user’s needs, a “micro-domain ontology” based personalized search strategy will be studied in this project. This ontology aims to model the user’s context into such a hierarchical structure so that with whose help the user’s intent can be disambiguated and the domain of interest can be identified. Based on the fact that terms are the basic building blocks for this ontology, we are particularly interested in term extraction in this context, especially by applying NLP techniques and a machine learning method that will leverage the minimal or zero prior knowledge. A micro-domain ontology is one of the core components in the target product for the partner organization. Our work will provide a POC (proof of concept) for the target product from this perspective.

Milan Tofiloski
Faculty Supervisor: 
Dr. Anoop sarkar
British Columbia