Exploring the Public Perception in Social Big Data: An Investigation in Mars Recall Scandal

Ma, Jie, Tse, Ying Kei, Zhang, Minghao, Sato, Yuji and Yu, Jin (2018) Exploring the Public Perception in Social Big Data: An Investigation in Mars Recall Scandal. In: 20th International Working Seminar on Production Economics, 19-23 February 2018, Innsbruck.

Preview

Text full paper for abstract043.pdf - Accepted Version Download (533kB) | Preview

Abstract

Social media has become a popular platform of interpersonal communication in which users can search for news and convey real-time information. Researching into social big data, such as Twitter, can be an effective way to identify public opinions and feelings in risk emergence, as it provides rich sources of data for opinion mining and sentiment analysis. This study aims to investigate and analyse the public perception towards the Mars and Snickers product recall scandal. The study proposes a comprehensive data analysis framework, and utilises the dataset formed of 10,930 Twitter messages over the span of 10-day following the product recall announcement made by Mars Inc., to gauge public attitudes and opinions. The study finds that the overall attitude of Twitter users towards the scandal was negative, and Snickers were the most mentioned product in the 10-day periods after the announcement of the recall. The data analysis highlights that the Tweet diffusion (retweeting) has positive associations with the number of followers and the use of hashtags, hence companies should pay more attention to users who have a large number of followers, as their tweets will be read by a great number of other Twitter users. The findings suggest effective methods for practitioners in crisis management (e.g., how to use social media to disseminate information). The study also presents a progressive tweet-mining framework that can serve as a tool in crisis management to classify the tweet topics, identify and analyse the sentiment and comprehend the changes of the public attitudes.

Actions (login required)

View Item

Downloads