Deep Learning based Opinion Mining on Throat Cancer Social Media Posts

Abstract

Twitter has become a popular platform for people to share their thoughts and opinions with the world. It allows users to post openly on any topic, giving them the freedom to express themselves without fear of judgment or censorship including those relevant to throat cancer. Twitter sentiment analysis is an important tool for understanding the relative sentiment of the public for certain topics or ideas present on the platform. By using Natural Language Processing (NLP) techniques on millions of tweets, Sentiment Analysis determines how likely each tweet falls into a pre-defined positive or negative classification. The tweets will be classified into three categories using the Lexicon, CNN, LSTM, and CNN-LSTM: positive, neutral, and negative. This study examined the use of text tweets from Twitter as a source of data. Curated tweets from public accounts were utilized and a total of 30002 tweets were collected. The study suggests that the use of Lexicon, CNN, LSTM, and CNN-LSTM approaches can enhance accuracy when conducting a classification task. Through this process, 82% accuracy has been obtained with 24000 positive tweets and 6000 negative tweets.