Social networks and energy pricing : a sentiment analysis based approach
Loading...
Date
2023-09-20
Authors
Ανδρούτσος, Χρήστος
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The technological advancement and the spread of smart devices have made the
Internet and social networks an integral part of today’s mankind’s routine. Users
post daily and convey their opinions on a variety of topics, which generates an
enormous volume of data. Sentiment Analysis, a subfield of Artificial Intelligence,
enables the exploitation of this information by those who are directly interested.
In this thesis, we attempted to depict the response of Greek consumers to the
rise in energy prices. We utilized data (tweets) from the social network Twitter for
this purpose. Our analysis was based on the technique of sentiment analysis employing
supervised machine learning and the Naive Bayes algorithm in particular.
We utilized a 10-Fold Cross Validation procedure to assess the precision of our
model. Our results indicate that our model can be used to predict the response
to an increase in the price of energy by analyzing user tweets on the subject.
Then, we entered the unknown data into the algorithm, which categorized 38.39%
of them as positive and 61.61% of them as negative, demonstrating the negative
consumer response. Lastly, in many tweets, users propose measures to address
the rise in energy prices, primarily the reduction of taxes, which appears in 8.25%
of all tweets, followed by the use of domestic sources of lignite to satiate demand
electricity, which appears in 1.56%.
Description
Keywords
Sentiment analysis, Social media, Behavioral economics