Abstract:
Information retrieval (IR) has become an important application in today's computer world because of the great increase in the amount of web-based documents and the widespread use of the Internet. However, the classical "bag of words" approach no longer meets user expectations adequately. In this context, the use of natural language processing (NLP) techniques comes into mind. In this thesis, we investigate the question of whether NLP techniques can improve the effectiveness of information retrieval in Turkish. We implemented a linguistically motivated information retrieval system, called TURNA (TUrkish information Retrieval engine based on Natural language Analysis). The system uses knowledge of three different levels of natural language processing in document and query processing: morphological, syntactical and lexico-semantical levels. Different combinations of these NLP techniques are tested on a set of Turkish documents and queries. The results are evaluated in terms of precision and recall. It is shown that natural language processing techniques, especially stemming and the use of syntactical head-modifier pairs, can improve information retrieval effectiveness in Turkish.
