Analisis Sentiment Twitter Berbasis Grid Search Algorithm (GSA) Dengan Metode Support Vector Machine (SVM)

Abstract

Twitter is a social networking service that has undergone tremendous growth and is gaining worldwide popularity at an accelerated rate. Twitter allows for the expression of unbiased thoughts on a variety of issues and can assist businesses in providing public feedback on well-known brands and items. Twitter is having trouble with good and negative answers. Researchers evaluated English-language tweets to determine the proportion of positive and negative replies to popular companies and items. This study will explore Twitter sentiment analysis utilizing the Grid Search Algorithm (GSA) and the support vector machine (SVM) technique. GSA is utilized by the feature selection model to optimize the classification procedure. In this work, training data and testing data are required to do sentiment analysis. Sanders Twitter 0.2 utilizes a dataset consisting of tweets retrieved from Twitter using the search terms @apple, #google, #microsoft, and #twitter. The collected dataset was manually annotated and included 654 negatives, 570 positives, 2503 neutrals, and 1786 irrelevant entries. Data are loaded, tokenized, weighted, preprocessed, filtered, and classified to conduct a sentiment analysis. The application's sentiment analysis achieved a degree of accuracy of up to 79% based on testing. The ratio of neutral and bad tweets on data sandboxes tends to be greater than the percentage of positive tweets, hence optimization rather than accuracy is obtained.