Web based customizable vertical search engine for turkish documents

From the beginning Internet has been expanding constantly, transformed into an unprecedented sea of knowledge and Internet usage has become a part of daily life. As Internet is getting bigger, finding needed information among this huge source of information is becoming more important. At this point search engines has taken the burden and made information access easier. Despite their success on meeting general information needs, search engines - indexing whole web - which are called horizontal search engines, couldn?t repeat the same success on meeting information needs concerning specific interest areas. Except few languages most web content is created, like English, horizontal search engines don?t consider the characteristic of content language. To overcome the difficulties horizontal search engines facing meeting specific information needs use of vertical search engines focused to some specific area, using special indexing, searching and ranking algorithms suitable for specific area, came into question.In this thesis, we aimed to develop a user customizable web based vertical search engine which considers characteristics of Turkish language. It is aimed to index Turkish documents more efficiently by processing the Turkish characters found in pages correctly, identifying the document language and stemming tokenized text. We tested both sub components of search engine and whole vertical search engine and saw increased information retrieval performance.Because our vertical search engine has an extendable architecture all sub components can be extended and customized if needed.Keywords: Vertical Search Engines, Turkish Information Retrieval, Heritrix,Lucene, Language Identification