Analyzing Public Concern Responses for Formulating Ordinances and Laws using Sentiment Analysis through VADER Application

Abstract

Purpose – This paper aimed to develop a system that applies VADER Sentiment Analysis to tweets collected using a developed twitter scraper tool to identify the insights of public responses based on their tweets on certain government services rendered to them thus providing legislators of the province of Laguna an additional tool in writing future legislations.

Method – This study may serve as an additional tool to the Sangguniang Panlalawigan of Laguna in identifying sentiments of the public in terms of government services that are rendered and lack thereof based on the collected tweets written in Tagalog, English or Taglish (Tagalog and English). Data collected through the Twitter scraper tool are preprocessed taking into consideration the special characters that also have impact on scoring sentiments, emojis, and emoticons. The compound score is computed by normalizing the sum of the polarity scores for each tweet.

Results – Aside from a tabular visualization of VADER’s results, the system also provides graphical representation of the evaluation result with the percentage of positive neutral and negative tweets. Based on the result of the testing and evaluation, the VADER model is 80.71% accurate and had an F-score of 84.33%.

Conclusion – The reports generated from the system be utilized to serve as potentially additional basis for legislators of the province of Laguna in writing legislations such as resolutions and ordinances based on the sentiment or voice of the community.

Recommendations – It is recommended to collaborate with linguists to develop a native language of VADER’s lexicon to improve the accuracy of the sentiment scores.