Android based sentiment analysis application with lexical similarity method

In this study, a mobile application which does sentiment analysis with lexical similarity method was performed in Android based mobile devices. Sentiment dictionary used in the sentiment analysis with lexical similarity method was developed using the WordNet lexical database method which expresses the scope as upper meaning (hypernym), inclusion as sub meaning (hyponym), being at the same level as coordination (coordinate) and the Resnik WordNet similarity measure method. English words to Turkish words were translated by using like Google, Yandex dictionary applications. Positive, negative labeled film and hotel comments used to make datasets. Sentiment analysis was made with sentiment dictionary and datasets. Datasets as normal and without stop words forms were used in sentiment analysis. Sentiment analysis results were evaluated by accuracy and F-Measure performance measures. Sentiment analysis success rate is about %56. Java platform and libraries were used in the process of generating sentiment dictionary and using it in sentiment analysis. Android mobil application developed in the Android Studio platform with Turkish sentiment dictionary.