Sentiment Analysis using Optimised Feature Sets in Different Facebook/Twitter Dataset Domains with Big Data

Nowadays, in many real-life applications, sentiment analysis plays a vital role in the automatic prediction of human activities, especially on online social networks (OSNs). Since the last decade, the research on opinion mining and sentiment analysis has been growing with the increase in volume of online reviews available over the social media networks, such as Facebook OSNs. Sentiment analysis falls under the data mining domain research problem. Sentiment analysis is a type of text mining process to determine subjective attitude, such as sentiments from written texts, and hence has become a main research interest in domains of natural language processing and data mining. The main task of sentiment analysis is to classify human sentiments with the objective of classifying the sentiment or emotion of end users based on their specific text on the OSNs. Several research approaches have been designed for sentiment analysis, in which the factors of accuracy, efficiency and speed have been used to evaluate the effectiveness of sentiment analysis methods. The Map-Reduce framework under the domain of big data is used to minimise the speed of execution and efficiency recently with many data mining methods. The sentiment analysis of messages on Facebook OSNs is more challenging compared with those of other sites because of the misspellings and slang words in their Twitter dataset. In this study, the different solutions that have been recently presented are discussed in detail. Then, a new approach for sentiment analysis based on hybrid feature extraction and multi-class support vector machine methods is proposed. The algorithms have been designed using big data techniques in view of optimising the performance of sentiment analysis. © 2022 Iraqi Journal for Computer Science and Mathematics. All rights reserved.