Page 119 - Kaleidoscope Academic Conference Proceedings 2024
P. 119
A FRAMEWORK FOR FINE GRAINED SENTIMENT ANALYSIS ON CODE-MIXED
LANGUAGE FOR SOCIAL MEDIA USER BEHAVIOURS
2
1
Anand, Tank ; Pratik, Vanjara
1 Atmiya University – Rajkot, India
2 Shree M. P. Shah Commerce College – Surendranagar, India
ABSTRACT
Extraction and comprehension of human dynamics,
Here, we provide a framework that discovers sentiments including actions, trends, attitudes, and emotions, are made
from social media platforms, assesses, and transforms them easier by sentiment analysis of subjective data [3]. Through
into meaningful data. Social media is changing people's forums, blogs, wikis, social networks, and other online tools,
attitudes and habits, which in turn is influencing their millions of people communicate their thoughts and feelings.
choices. Attempting to keep an eye on social networking Sentiment Analysis (SA) examines a user’s thoughts,
activity is a useful tool for tracking consumer attitude about attitudes, perspectives, ideas, beliefs, remarks, requests,
products and firms and gauging loyalty from consumers. The inquiries, and preferences toward various things such as
framework can be the next natural area for branding based services, issues, people, products, events, subjects,
on the polarities on the internet and social media. We present organizations, and their characteristics, based more on
a dynamic solution method for sentiment analysis using the emotion than on logic in the form of writing. It determines
classification of interpersonal data sources. To evaluate the the general mood of the writer for a text, which could include
caliber of social information services, we also introduce a speeches, product evaluations, blog entries, online forums,
brand-new quality model. We utilize public comments, posts database sources, social media information, and papers.
through social media as an inspiring case study. Specifically, Depending on the circumstances, it typically consists of three
to pinpoint the comments and posts we concentrate on the components: 1) Opinions or emotions, 2) Subject, 3) Opinion
spatiotemporal characteristics of the attitudes expressed by Holder [4].
social media users. On datasets from the real world,
experiments are carried out. Our suggested model’s Sentiment analysis of subjective datasets facilitates the
performance is preliminary demonstrated by performance extraction and understanding of human dynamics, including
evaluation matrix. behaviors, patterns, attitudes, and emotions [5]. A lot of
noise, or superfluous and useless material, is frequently
Keywords – Sentiment Analysis, social media, Artificial present in social Datasets. Furthermore, there are many kinds
Intelligence, Machine Learning, Natural Language of social media information services available on the internet.
Processing On the internet, there are numerous data providers that have
distinct data characteristics, such as size, quality, and so on.
1. INTRODUCTION Thus, every dataset needs a different approach for extracting
useful information. It takes time to use different tools, and
Societies use their sentiments to convey their opinions and the perspectives of the data from social sensors are not
behaves as per experience. Sentiment analysis is the term always consistent. [6].
used to describe the analysis of these kinds of perspectives.
Social networks, or social networking sites, like Facebook, Our study presents a system called "Fine Grained Sentiment
Twitter, and others, have become a free community for Analysis for Code-Mixed Languages" (FGSACML), which
dataset [1]. When there is an event, a lot of people use social gathers sentiments for code-mixed language from multiple
media services to quickly create and share data i.e. Twitter, social media sites, analyzes and transforms them into
Facebook, LinkedIn, etc. [2]. This data generated by social meaningful data, and then returns the data in the form of a
media users has many useful information as metadata: 1) It polarity. Using different characteristics of the data from the
has subjective information as Opinion. 2) It includes social media user, we analyzed social information like
spatiotemporal information as Behavior, among others. Positive, Negative and Neutral. Conventional techniques for
sentiment evaluation also cover the numerous attributes of
media information like comments, posts and review.
Nevertheless, this framework transcends polarities and is not
just based on sentiment. Two primary methods can be used
978-92-61-39091-4/CFP2268P @ITU 2024 – 75 – Kaleidoscope