Sentiment analysis is a very popular applicationarea of text mining and machine learning. The popular methodsinclude support vector machine, naive bayes, decision trees, anddeep neural networks. However, these methods generally belongto discriminative learning, which aims to distinguish one classfrom others with a clear-cut outcome, under the presence ofground truth. In the context of text classification, instances arenaturally fuzzy (can be multilabeled in some application areas)and thus are not considered clear-cut, especially given the factthat labels assigned to sentiment in text represent an agreed levelof subjective opinion for multiple human annotators rather thanindisputable ground truth. This has motivated researchers todevelop fuzzy methods, which typically train classifiers throughgenerative learning, i.e., a fuzzy classifier is used to measure thedegree to which an instance belongs to each class. Traditionalfuzzy methods typically involve generation of a single fuzzyclassifier and employ a fixed rule of defuzzification outputting theclass with the maximum membership degree. The use of a singlefuzzy classifier with the above-fixed rule of defuzzification is likelyto get the classifier encountering the text ambiguity situation onsentiment data, i.e., an instance may obtain equal membershipdegrees to both the positive and negative classes. In this paper,we focus on cyberhate classification, since the spread of hatespeech via social media can have disruptive impacts on socialcohesion and lead to regional and community tensions. Automaticdetection of cyberhate has thus become a priority research area.In particular, we propose a modified fuzzy approach with twostage training for dealing with text ambiguity and classifyingfour types of hate speech, namely, religion, race, disability, andsexual orientation—and compare its performance with thosepopular methods as well as some existing fuzzy approaches, whilethe features are prepared through the bag-of-words and wordembedding feature extraction methods alongside the correlationbased feature subset selection method. The experimental resultsshow that the proposed fuzzy method outperforms the othermethods in most cases.
To View the Abstract Contents
Now it is Your Time to Shine.
Great careers Start Here.
We Guide you to Every Step
Success! You're Awesome
Thank you for filling out your information!
We’ve sent you an email with your Final Year Project PPT file download link at the email address you provided. Please enjoy, and let us know if there’s anything else we can help you with.
To know more details Call 900 31 31 555
The WISEN Team