Text Classification Using Improving Term Frequency with Gaussian and Multilayer Perceptron Techniques

This paper proposes combining the term frequency with the Gaussian technique and the multilayer perceptron technique for text sentiment analysis using two datasets: tweet customer sentiments serving U.S. airlines and Amazon product sentiment reviews. The bag of words, term frequency-inverse document frequency, term frequency-inverse corpus document frequency, term frequency-inverse gravity moment, and term frequency with Gaussian are investigated in this study. The experimental results show that improving term frequency with Gaussian for term weighting in text sentiment analysis from tweets of customer sentiments serving U.S. airlines, with a higher F-score than previous techniques and an 8% increase in F-score from the original term frequency with Gaussian technique. Also, the term frequency with Gaussian technique was enhanced for term weighting in text sentiment analysis using Amazon product review scraping, with a higher F-score than the previous techniques and had a 7% increase in F-score compared to the original term frequency with Gaussian technique.