Opinion Mining on Twitter Data using Unsupervised Learning Technique

Abstract

Social media is one of the biggest forums to express opinions. Sentiment analysis is the procedure by which information is extracted from the opinions, appraisal and emotions of people in regards to entities, events and their attributes. Sentiment analysis is also known as opinion mining. Opinion mining is to analyze and cluster the user generated data like reviews, blogs, comments, articles etc. These data find its way on social networking sites like twitter, facebook etc. Twitter has provided a very gigantic space for prediction of consumer brands, movie reviews, democratic electoral events, stock market, and popularity of celebrities.The main objective of opinion mining is to cluster the tweets into positive and negative clusters. An earlier work is based on supervised machine learning (Naïve bayes, maximum entropy classification and support vector machines). The proposed work is able to collect information from social networking sites like Twitter and the same is used for sentiment analysis. The processed meaningful tweets are cluster into two different clusters positive and negative using unsupervised machine learning technique such as spectral clustering. Manual analysis of such large number of tweets is impossible. So the automated approach of unsupervised learning as spectral clustering is used. The results are also visualized using scatter plot graph and hierarchical graph.

Effects of using wordnet and spelling checker on classification methods in sentiment analysis for datasets using Bahasa

“…Dey et al [13] that explain Naive Bayes is better than other classification methods for hotel review case studies, Suhariyanto et al [14] that explain support-vector machine (SVM) is better than other classification methods for detection fake movie [15], decition tree is better than Naïve Bayes and k-NN [16], k-NN is better than Naïve Bayes [17], k-nearest neighbor classifier is transparent, consistent, straightforward, simple to understand, high tendency to possess desirable qualities and easy to implement than most other machine learning techniques [18]. The process of extracting data sets on social media is more suitable for using unsupervised machine learning than supervised learning [19], subversion (SVN) has the lowest accuracy than k-NN and Naive Bayes [20], sentiment analysis is widely applied in fields related to opinions describing satisfaction or dissatisfaction such as case studies of tourist reviewers, film reviewers and the like [21], [22]. It is these various case studies that produce many conclusions, because each of these cases also uses various datasets, both training data and testing data, so that they have the potential to produce various kinds of conclusions.…”

Unsupervised Twitter Sentiment Analysis on The Revision of Indonesian Code Law and the Anti-Corruption Law using Combination Method of Opinion Word and Agglomerative Hierarchical Clustering

“…Muqtar Unnisa, et al [3] from Darussalam Hyderabad TS. Based on the results of research that has been done, the first research is measuring public opinion with a reasonable assessment in film reviews with a machine learning algorithm based on spectral clustering for sentiment analysis.…”

A Calibration Method for the Evaluation of Sentiment Analysis

“…Similarly, unsupervised learning methods directly produce positive or negative labelled outputs using different techniques and algorithms such as k- means, TF-IDF and PMI-IR algorithms (Turney, 2002;Zagibalov and Carroll, 2008;Unnisa et al, 2016), but again without a clear interpretation of the classes identified. Lexicon-based approaches (as well as some unsupervised learning methods methods such as Turney (2002)) have proceeded by calculating the semantic orientation (a numerical score) and deciding the polarity of the document depending on its sign and the sentiment strength based on its magnitude.…”