Emotion recognition in election day tweets using optimised kernel extreme learning machine classifier

ABSTRACT

Today, microblogging has turned into a very well-known specialised device among web clients. A huge number of clients share ideas on various aspects of daily life. In this manner these sites are rich sources of information utilised with the goal of sentiment investigation. This investigation is essentially valuable as well as challenging since everybody consistently need to know the views of existing clients about an item or service. In this article, the most prominent microblogging platform ‘Twitter’ is selected for the investigation of sentiment classification and mining opinions. This is helpful for shoppers who need to enquire about items before purchase, or organisations that need to screen the open assumption of their brands. A unique approach is presented for classifying the feeling of Political Twitter messages into Happy, Extremely Happy, Sad, Extremely Sad or else Neutral. There are numerous devices that give computerised opinion investigation. The dataset used here is the Twitter corpus Dataset named 2016 Political Election day tweets gathered using the Twitty and Simulation apparatus utilised in this examination is PYTHON. Utilising the corpus, sentiment classifier named Kernel Extreme Learning Machine (KELM) optimised by Salp Swarm Algorithm (SSA) determines the class of political tweets as Happy, Extremely Happy, Sad, Extremely Sad and Neutral Sentiments. This work mainly focus on English tweets. The proposed KELM classifier performance is compared with existing classifier approaches and noted with highest accuracy which defines the effective nature of the classifier. Moreover, test evaluation demonstrate that the proposed procedure is effective and performs superior to recently proposed works.

KEYWORDS:

• Sentiment analysis
• Twitter data
• natural language processing
• text classification
• emotion recognition

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.https://authorservices.taylorandfrancis.com/data-sharing/share-your-data/data-availability-statements/

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the No Funding [No].