Many approaches to sentiment analysis benefit from polarity lexicons. Existing methods proposed for building such lexicons can be grouped into two categories: (1) Lexicon based approaches which use lexicons such as dictionaries and WordNet, and (2) Corpus based approaches which use a large corpus to extract semantic relations among words. Adjectives play an important role in polarity lexicons because they are better polarity estimators compared to other parts of speech. Among natural languages, Turkish, similar to other non-English languages suffers from the shortage of polarity resources. In this work, a hybrid approach is proposed for building adjective polarity lexicon, which is experimented on Turkish combines both lexicon based and corpus based methods. The obtained classification accuracies in classifying adjectives as positive, negative, or neutral, range from 71% to 91%.
