Automatic Sentiment Analysis Model Creation using Multi Kernel Improved Extreme Learning Machine
Srinidhi B. S.1, Suchithra R.2

1B. S. Srinidhi, Scholar, Department of MCA, Jain Deemed to be University, Bangalore, India.
2R. Suchithra, Director, Department of MCA, Jain Deemed to be University, Bangalore, India.
Manuscript received on March 12, 2020. | Revised Manuscript received on March 25, 2020. | Manuscript published on March 30, 2020. | PP: 2727-2735 | Volume-8 Issue-6, March 2020. | Retrieval Number: E6842018520/2020©BEIESP | DOI: 10.35940/ijrte.E6842.038620

Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: Recent research activities related to opinion mining, sentiment analysis and emotion detection from natural language texts are all under the umbrella of affective computation. There is now a huge amount of textual information on social media (for example, forums, blogs, and social media) about consumers’ ideas about buying products and service experiences. Sentiment analysis or opinion mining is part of an investigation that analyzes people’s thoughts and feelings from written text available online. In this paper, this work present a comprehensive experiment to evaluate the effectiveness of psychological and linguistic features in emotion classification. In this scheme, we used five broad categories of LIWC (namely, psychological processes, linguistic processes, punctuation, spoken categories and personal concerns) as feature sets. Five types of LIWCs and their group combinations were considered in the experimental analysis. To understand the predictive performance of various aspects of the engineering scheme, five controlled learning algorithms (namely, Naïve Bayes, support vector machines, Extreme Learning Machine, Kernel Extreme Learning Machine, Multi Kernel Extreme Learning Machine) and proposed Multi Kernel Improved Extreme Learning Machine (MKIELM) are used. Experimental results show that the ensemble feature sets provides a higher predictive effect than the individual set.
Keywords: Natural Language Processing (NLP), Emotion Recognition, Social Media Analysis, Sentiment Analysis, Psychological Feature Sets, Twitter, ELM And MKIELM.
Scope of the Article: Big Data Analytics for Social Networking using IoT.