Document Type

Project

Lead Author Type

CIS Masters Student

Advisors

Dr. Jonathan Leidig, jonathan.leidig@gvsu.edu

Embargo Period

8-12-2014

Abstract

An ontology is a representation of knowledge as hierarchies of concepts within domain, using a shared vocabulary to denote the types, properties and inter-relationships of those concepts [1][2]. Ontologies are often equated with classification of hierarchies of classes, class definitions, and the relations, but ontologies need not be limited to these forms. Ontologies are also not limited to conservative definitions, i.e., in the traditional logic sense that only introduce terminology and do not add any knowledge about the world (Enderton, 1972). To specify a conceptualization, axioms need to be proposed that constrain interpretation of defined terms [3].

Ontologies are frameworks for organizing information and are collections of URIs. It is a systematic arrangement of all important categories of objects and concepts within a particular field and relationship between them. Search engines are commonly used for information retrieval from web.

The ontology based personalized search engine (OPSE) captures the user’s priorities in the form of concepts by mining through the data which has been previously clicked by them. Search results need to be provided according to user profile and user interest so that highly relevant search data is provided to the user. In order to do this, user profiles need to be maintained. Location information is important for searching data; OPSE needs to classify concepts into content concepts and location concepts. User locations (gathered during user registration) are used to supplement the location concepts in OPSE. Ontology based user profiles are used to organize user preferences and adapt personalized ranking function in order for relevant documents to be retrieved according to a suitable ranking. A client-server architecture is used for design of ontology based personalized search engine. The design involves in collecting and storing client clickthrough data. Functionalities such as re-ranking and concept extraction can be performed at the server side of personalized search engine. As an additional requirement, we can address the privacy issue by restricting the information in the user profile exposed to the personalized mobile search engine server with some privacy parameters. The Prototype of OPSE will be developed on the web platform. Ontology based personalized search engines can significantly improve the precision of results.

Share

COinS