Authors:
Aman Agarwal
and
Veena Bansal
Affiliation:
Indian Institute of Technology Kanpur, India
Keyword(s):
Sentiment Analysis, Unstructured Data, Campaign Management.
Abstract:
Purpose: Winning an election requires more than a good and appealing manifesto. The purpose of this paper is to establish that content from the social media provides useful insights and can be used to manage an election campaign provided the right content is properly analyzed. Information such as frequency of mentions, sentiments of the mentions and demography is obtained analysis. This information provides insights into the demography of supporters, topics that are most talked about revealing their importance to the voters, sentiments of voters. Design/Methodology/Approach: We analyzed 25000 documents from twitter, forums, reviews, Facebook pages, blogs etc. over a period of 12 months in three states of India using Watson Analytics for Social Media (WASM) of IBM. We used ETL (extract, transform and load) utility of WASM to fetch the documents for our chosen themes, topics, dates and sources. WASM deploys deep learning to perform sentiment analysis. Findings: We found that social med
ia content analysis provides useful insight that goes beyond general perception and can be used for managing a campaign. Originality/Value: There have been many efforts where researchers are trying to predict election results based on social media analysis. However, these efforts have been criticized as predicting election results is a very complex problem. We, in this work, have shown that social media content can definitely help in gaining a clear understanding of the sentiments of voters.
(More)