Abstract:
"More and more people use emoticons in their texts to communicate their feelings or remember
their comments. Previous machine learning algorithms mainly focused on categorizing text,
emoticons, or images and ignored emoticons with text, resulting in many emotions being
skipped. The study provided a sentiment analysis algorithm and approach that used text and
emoticons. Both data forms were evaluated in this study using machine learning to detect
sentiments from Instagram comment data using numerous features such as TFIDF, N-gram,
and emoticon lexicons. This study shows that when using emoticons, the associated sentiment
outperforms the sentiment represented by the analysis of text data."
