Social Media Text Analysis Based on Soft Computing

Abstract

Now a days, communication through social media platforms is a backbone of our newlineday to day life. Social media users feel free to share formal as well as informal newlinecommunication including views, thoughts, emotions, opinions and sentiments. newlineEnormous data is moving through social webs, this huge data need attention. As newlinepeople communicate informally, abnormal and harmful content are there. This is newlinethe responsibility of social media providers to maintain healthy communication newlineamong its users. People feel comfortable to express their high emotions/sentiments newlinein their local languages. Thus offensive, abusive and hate content are mostly newlineavailable in low resource languages including code-mix languages as well as newlineEnglish languages. Natural language processing become more important as 94 newlinezettabytes1 newlineof internet data are produced and consumed by internet users in 2022 as newlinemost of the data are in natural language. newlineOn social media, people communicate through natural language which is uncertain, newlineimprecise and ambiguous in nature. It doesn t not has any grammatical form in newlineaddition to this it also involves abusive words, slang words, code-mix languages, newlinesarcasm, metaphor, ambiguous and emojis. Detection of offensive, hate and abusive newlinetext is a challenging research area due to all these issues. To provide healthy newlinecommunication, researchers are working for hate speech detection, abusive text newlinedetection, revenge text detection, suicide and depression detection and aspect based newlinesentiment analysis. For hate speech detection researchers are not provide newlineunsupervised, lexicon based, and knowledge based work for low resource newlinelanguages. Present state of the art are not able to provide contextual analysis of newlineabusive, aggressive and misogynistic text. Many times social media revenge text newlineare in form of long sentences where semantic relation dissolves between tokens. newlineDue to that, researchers did not provide any attention towards identifying the users newlinespreading revenge. Along with that, aspect sentiment aggregation research newlinecomputation ...