Date of Award


Publication Type

Master Thesis

Degree Name



Computer Science


Applied sciences


Ziad Kobti




Many of the problems in natural language processing (NLP) or information retrieval (IR) stem from the rich expressive power in natural language. The use of concept search to overcome the limitations of keyword search has been put forward as one of the motivations of the Semantic Web since its emergence in the late 90's. A lot of efforts have been made to adapt concept search principles for improving the performance of information retrieval systems. However most approaches are designed for data repositories which contain a large number of items, each with rich information. We propose a knowledge based concept search method that is particularly designed for the data repository with limited information items by narrowing down a query into one concept. Also, a framework adapting this method is proposed to solve the practical problem about how to extract the information from a specification document into an existing unstructured database. As an important part of the framework, a concept selection method using genetic algorithms and semantic distance is proposed to filter the meaningless or less important information in query generation and matching process. An application development process in mould engineering domain is introduced as a case study to show how to use this framework. The experiment results show our proposed concept search method performs better than classical keyword based search especially for the documents with ambiguous words and the concept selection method has a potential to further improve this concept search method.