Semi-supervised Language-independent Sentiment Analysis

Abstract

The rapid growth of the internet has changed the way people do business and the way they communicate. With billions of active [1] users every day from all the globe, the internet has given a globalized dimension to the world we live in. Millions of posts, blogs, reviews, news and books are published publicly every day that cover every aspect and detail of our life. This abundance of information creates which is called the information age [2] and with a smart analysis of the available data, a lot of valuable information could be extracted. The available data may contain subjective information i.e. opinions, sentiments or objective information i.e. facts. Sentiment analysis (SA) or opinion mining is the field that automates the process of de- tecting the subjective information conveyed in texts i.e. opinions, feelings and emotions then classifies them into positive, negative or neutral based on the texts‘ features. SA plays an important role industry and has gained the interest of researchers in different application such as stock market prediction, box office prediction, business analytics and marketing Intelligence [3]. Twitter1 is one of the most popular social media platforms, with 100 million daily active users and 500 million daily tweets in 2018 [4], twitter has attracted many researchers to excavate the valuable information included in tweets. Different researches were provided to overcome the various challenges in data such as the unstructured format of data, subjectivity of texts, using different languages any many more.