A New Approach to Analysis of Public Sentiment on Padma Bridge in Bangla Text

Abstract

Opinion mining or emotion analysis of people is generally called sentiment analysis. Sentiment analysis is necessary for getting customer or user notions about any products or services. This research paper investigated and recorded the public sentiment on Padma Bridge. As a result, the Government of Bangladesh can make decisions easily on future mega-construction projects based on the recorded results. Padma bridge is a mega construction event for Bangladesh due to the big budget and banned World Bank loans. Bangladeshi people express their feelings, suggestions, opinions, and thoughts about the Padma Bridge project on Facebook, YouTube and other social media. The main focus of this paper is sentiment analysis of people’s reactions to Padma bridge is based on the Bangla comment dataset. We have collected more than 15K data which has two types of sentiment: Positive and Negative. Then we used three machine learning models (SVM, RF, LSVC) and one deep learning model (LSTM) for sentiment analysis. In our proposed system, we used an innovative voting method that can count and compare the sentiments produced by the mentioned ML and DL models. Finally, our model makes decision-based on the maximum voting results. This paper concludes that the voting method technique improves the accuracy by around 7.5% compared to every single ML and DL model.