
Structured Learning Based Turkish Sentiment Analysis
Author(s) -
Oguz Ulgen,
Arif Selçuk Öǧrenci
Publication year - 2019
Publication title -
akıllı sistemler ve uygulamaları dergisi
Language(s) - English
Resource type - Journals
ISSN - 2667-6893
DOI - 10.54856/jiswa.201912071
Subject(s) - sentiment analysis , turkish , computer science , the internet , social media , artificial intelligence , data science , world wide web , linguistics , philosophy
Sentiment analysis is highly popular topic to identify people's opinions through the social media, forums and other websites. There are an abundance of opinions on internet and analysing those opinions would have many benefits for both private and public sectors. Research has evolved looking on tweets for mining opinions and for the classification of the tweets as positive, negative or neutral in its sentiment. In this research, Turkish tweets are used for sentiment extraction where a two layer neural network is used as the pattern recognition system. The supervised training of this system is based on structured learning. As a conclusion, structured learning seems to be helpful in pattern recognition to classify tweets and mining the opinions. However, it is evident that further research in data processing and training methodology is necessary to obtain reliable sentiment analysis results.