Open AccessSENTIMENT ANALYSIS OF ENGLISH TWEETS USING BIGRAM COLLOCATION
Author(s) -
Sumaya I. Moyeen,
Md. Sadiqur Rahman Mabud,
Zannatun Nayem,
Md. Al Mamun
Publication year - 2021
Publication title -
epra international journal of research and development
Language(s) - English
Resource type - Journals
ISSN - 2455-7838
DOI - 10.36713/epra8524
Subject(s) - bigram , sentiment analysis , naive bayes classifier , support vector machine , computer science , artificial intelligence , machine learning , principle of maximum entropy , microblogging , classifier (uml) , natural language processing , entropy (arrow of time) , python (programming language) , collocation (remote sensing) , social media , world wide web , physics , trigram , quantum mechanics , operating system
Community and portal websites like Twitter, Facebook, Tumbler, Instagram, and LinkedIn etc. have significant impact in our day-to-day life. One of the most popular micro-blogging platforms is twitter that can provide a huge amount of data which in future can be used for various applications of opinion mining like predictions, reviews, elections, marketing etc. The users use this platform to share their views, express sentiments on various events of their daily life. Previously, many researchers have worked with twitter sentiment analysis and compared various classifiers and got the accuracy below 82%. In this work for classifying tweets into sentiments, we have used various classifiers such as Naïve Bayes, Support Vector Machine and Maximum Entropy that segregate the positive and negative tweets. Using Bigram Collocation with classifiers, we’ve acquired 88.42% accuracy.KEYWORDS: Twitter; Sentiment Classification; Machine Learning; NLTK; Python; Naïve Bayes; Support Vector Machine (SVM); Maximum Entropy