Manuscript Title:

ENHANCED TWITTER SENTIMENT ANALYSIS USING HYBRID CLASSIFICATION METHODS AND RESULT ANALYSIS

Author:

Priyanka Tyagi, Dinesh Javalkar, Sudeshna Chakraborty

DOI Number:

DOI:10.17605/OSF.IO/2NVJK

Published : 2021-06-23

About the author(s)

1. Priyanka Tyagi - Research Scholar, Lingayas Vidyapeeth, Faridabad.
2. Dinesh Javalkar - Asst. Professor Lingayas Vidyapeeth, Faridabad.
3. Sudeshna Chakraborty - School of Engineering and Technology, Sharda University.

Full Text : PDF

Abstract

The utilization of social media, in the course of recent years, has heightened hugely.Social media has framed a stage for the accessibility of plentiful information. A great many individuals express their discernments through social media. Sentiment Analysis (SA) of such perspectives and discernments is considerable to gauge public opinion on a particular/explicit topic of concern. Twitter is a microblogging webpage wherein clients can post updates (tweets) to companions (supporters).This paper proposes an instrument for extricating the suppositions from the tweets posted on Twitter. Tweets can be delegated positive, nonpartisan or negative. This model ends up being profoundly compelling and exact in the investigation of sentiments. This paper presents a crossover approach of utilizing Subterranean insect Settlement Enhancement and Molecule Multitude Improvement with classifiers. For each tweet, pre-handling will be finished by performing different cycles, for example tokenization; expulsion of stop-words, stemming etc. Besides, highlights are removed by the Uni-gram include extraction procedure. Grouping is completed by a Help Vector machine (SVM) classifier with Subterranean insect State Enhancement and Molecule Multitude Improvement to advance the arrangement execution. For solving ACSO, this work presents a hybrid classification based on ant colony optimization (ACO) and particle swarm optimization (PSO). In addition, the ACO framework includes a pheromone disruption method for dealing with pheromone stagnation. The execution showed that the new half breed slant arrangement had the option to further develop the precision execution. The productivity of the proposed framework was approved on the Creep Tweet dataset.


Keywords

Sentiment Mining, Naïve Bayes (NB), Support Vector Machine (SVM). k-nearest neighbor (KNN), hybrid classification method, Ant Colony Optimization (ACO),PSO , Sentiment analysis