文章基本信息

标题：A COMPARISON OF ML APPROACHES ON SENTIMENT ANALYSIS, BASED ON ONTOLOGIES, SARCASM AND SUBJECTIVITY DETECTIONS IN THE CASE STUDY OF US ELECTIONS
本地全文：下载
作者：IHAB MOUDHICH ; ABDELHADI FENNAN
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：5
语种：English
出版社：Journal of Theoretical and Applied
摘要：Nowadays, Twitter has become one of the most excellent tools that give people the power to express their emotions. And also, to interact with other ordinary or political people. According to The Verge, as known as the American technology news website, more than 166 million users are using Twitter every day; this thing made Twitter one of the largest news sources and one of the places where most of the politicians publish their opinions or their thoughts. Sentiment analysis or opinion mining is a method that is used to understand the user's behavior based on their feelings in a given text, which can help to get a global idea of the expected outcomes of USA elections. Our research is based on extracting the data and analyzing tweets' sentiment to predict the USA elections results. As we know, most Americans use Twitter to interact with each other to explain their opinions and thoughts about the subjects related to their country. Also, the hashtag system on Twitter makes it easy to help people to interact and go viral.We also included other variables to make a significant comparison of our results, such as detecting sarcasm and subjectivity in a tweet. Also, we used two machine learning approaches: First known as Long Short-Term Memory (LSTM). The second is Bidirectional Encoder Representations from Transformers (BERT). In this work, we used more than 500,000 tweets to get a significant result. Moreover, our developed Framework consists of 5 steps: First, collecting data based on ontologies that we defined. Second, text pre-processing to clean data. Third, predicting subjectivity and sarcasm in a tweet. Fourth applying the two cited approaches to get the sentiment. And the last one is visualizing and analyzing the results.
关键词：Sentiment Analysis;Machine Learning;Ontology;LSTM;BERT;Sarcasm Detection;Subjectivity Detection