A RULE BASED NOUN PHRASE CHUNKER FOR TURKISH

Authors : Kübra Adalı, A. Cüneyd Tantuğ

Pages : 17-25

View : 8 | Download : 10

Publication Date : 2014-12-21

Article Type : Research Paper

Abstract :In this paper, we presented a noun phrase chunker for Turkish as an agglutinative language. For finding noun phrases in Turkish sentences, we propose a rule based model which includes preprocessing part and a unit that applies the local grammatical rules to the output of the dependency parser. To the best of our knowledge, our model gives the first results on noun phrase chunking of Turkish sentences that is expected to find not only the basic noun phrase sentences but also the complex noun phrases including the relative clauses. We believe that on that sense, our model will be a good reference for future studies in this domain. We tested our model both on manually annotated data and the output version of the dependency parser. Our model gives the results with annotated data for full match 66.15\% and the partial match 76.79\% insert ignore into journalissuearticles values(for F1 results);. Using output of the dependency parser, the results are 47.91\% and 60.75\% for F1 results accordingly insert ignore into journalissuearticles values(for F1 results);
Keywords :