LSTM Derin Öğrenme Yaklaşımı ile Covid-19 Pandemi Sürecinde Twitter Verilerinden Duygu Analizi

Abstract

analiz ABSTRACT It is very important to understand people’s thoughts regarding social events occurring in the world and to make some inferences by analyzing these thoughts. With these analysis and inferences, various projects can be initiated and decision-making processes can be formed. One of the procedures used for these purposes is the sentiment analysis which is performed by classifying text with various computer algorithms. The methods used to perform sentiment analysis are generally categorized as dictionary-based methods and machine learning approaches. In this paper, a sentiment analysis study has been carried out by considering a number of frequently spoken terms on the Twitter social media platform regarding the coronavirus (Covid-19) pandemic, which has affected the world and is still ongoing. For this, some Turkish titles related to the subject were collected and sentiment analysis was conducted by classifying these titles as positive and negative thoughts. For this analysis, a system using a Long Short-Term Memory (LSTM) structure, which is one of the deep learning methods, was proposed. The proposed system was applied on the obtained data sets and a maximum 97% accuracy was achieved.