A method for part of speech tagging and finding type of sentence in Turkish

As the technology diffuses more and more in our lives, the usage and prominence of languages increases in the digital environment. For this reason, the main motivation of our study is to develop methods for understanding of Turkish sentences correctly in the digital world. In our work, it has been examined how to determine automatically the type of a sentence according to its meaning. For this; the performance of part of speech tagging and stemming task developed by Tarık Kışla has been improved and transformed into a platform independent structure. The types of the words in the sentence were first determined using the Hidden Markov Models and the Viterbi algorithm and then the stems and suffixes of the words were determined. With a rule-based method that uses this data, the type of sentence according to its meaning as positive sentence, negative sentence, positive interrogative sentence, negative interrogative sentence, has been labeled. Determining type of sentence according to its meaning, part of speech tagging and stemming methods were tested and the results have been compared with previous works on related topics.