COMPUTER AND INFORMATION SCIENCES - ISCIS 2004, PROCEEDINGS, vol.3280, pp.761-770, 2004 (SCI-Expanded)
This paper describes a preprocessor for Turkish text that involves various stages of lexical, morphological and multi-word construct processor for preprocessing Turkish text for various language engineering applications. We present the architecture of the system with special emphasis on how various kinds of collocations and other similar multi-word constructs are handled and present an evaluation from a test corpus.