Design and analysis of techniques for sarcasm detection using machine learning

In order to grasp individuals’ thoughts on various issues or events on social media, it is necessary to recognise both literal and metaphorical interpretations of words and phrases. Scanning social media for sarcastic remarks and tweets has lately gotten a lot of attention, especially because dismissive comments inside the manner of tweets sometimes include positive terms that indicate bad or undesirable attributes. Sarcasm may be used for a variety of objectives, including criticism and ridicule, depending on the situation. Even this, though, might be difficult for the average person to notice. Inorder to improve automated sentiment analysis obtained from various social media platforms and microblogging sites, thesarcastic restructuring technique is quite beneficial. A sentiment analysis refers to the stated views and preferences of internet users belonging to a certain community, which are then grouped together for further study. We present a pattern-based technique to identifying sarcasm that makes use of Twitter data. We suggest four groups of characteristics, each of which contains a significant amount of particular snarky. We utilised them to differentiate between sarcastic and non-sarcastic tweets. In addition, we examine each of the suggested feature sets and analyses the extra cost classes associated with them. Our findings also revealed that the use of lexical, sensible, periodicity, and part-of-speech tagging may improve the performance of SVM, but the use of both lexical and personal characteristics can improve the performance of many other machine learning algorithms, according to our findings. This paper also addressed the major difficulties that earlier researchers have encountered when attempting to forecast sarcastic tweets. Future academics or machine learning developers may find this information beneficial in addressing the key challenges associated with categorizing sarcastic postings in social media in the future.