Emotion classification of social media posts for estimating people’s reactions to communicated alert messages during crises

Abstract

One of the key factors influencing how people react to and behave during a crisis is their digital or non-digital social network, and the information they receive through this network. Publicly available online social media sites make it possible for crisis management organizations to use some of these experiences as input for their decision-making. We describe a methodology for collecting a large number of relevant tweets and annotating them with emotional labels. This methodology has been used for creating a training dataset consisting of manually annotated tweets from the Sandy hurricane. Those tweets have been utilized for building machine learning classifiers able to automatically classify new tweets. Results show that a support vector machine achieves the best results with about 60% accuracy on the multi-classification problem. This classifier has been used as a basis for constructing a decision support tool where emotional trends are visualized. To evaluate the tool, it has been successfully integrated with a pan-European alerting system, and demonstrated as part of a crisis management concept during a public event involving relevant stakeholders.

Informing crisis alerts using social media: Best practices and proof of concept

“…A detailed design concept for how a screening tool would potentially be used for increasing situational awareness during emergencies and crises, where data acquisition and data analysis were identified as two important parts of such a tool, was outlined by Johansson, Brynielsson, and Narganes Quijano (2012). Studies on tweet classification (Brynielsson, Johansson, Jonsson, & Westling, 2014;Brynielsson, Johansson, & Westling, 2013b) and a series of user-centred activities, involving emergency management personnel, aiming at understanding user needs, and informing the design of a social media screening tool (Brynielsson, Johansson, & Lindquist, 2013a) were conducted.…”

Systematic literature review of sentiment analysis on Twitter using soft computing techniques

“…Apple; Google; Microsoft, 27,29,34,44,55,73,75 consumer products like kindle, smart-phones, etc, 51,55,56,73,84 natural calamities, energy and environmental related, 36,46,77 cyber hatred, 47,62 entertainment 29,49,53,84 which includes tweets about music and movies, automotive or vehicles, 49,53 banking, 53 government or public campaign or public administration, 33,37,39,57,72,79 education or universities, 43,53,57,67 science or technology, 40,49,57 politics, 17,39,40,49,61,67,72,79,80,84 sports, 40,49,60 daily deals and discount, 56 trade or commercial services or business or financial...…”

“…Confusion matrix (CM) It is also called as error matrix and is used for evaluating 36 performances of the supervised learning classifiers based on the actual and the predicted classifications on a set of test data for which the true values are known. Each cell has fields like false negatives (FN), FP, TN, TP.…”

“…Author achieved P of 66.67% and R of 100% for the definitional tweets. In 2014, a methodology was developed by Brynielsson et al36 for selecting and collecting natural calamity related tweets and tagging them using the annotators like happiness, anger, fear, etc. They had focused on the applicability of the classifiers like SVM, and had observed that SVM outperformed NB with an improved A of about 60%.…”

Mining the Web for Sympathy: The Pussy Riot Case

“…The information obtained from such analysis could for example be useful for crisis management during a disaster. Sentiment analysis could then be used to monitor how the affected people are feeling and how they are responding to the help and the information they get [3], [4], [5], [6], [7]. The analysis can provide valuable information regarding what kind of help that would be the most useful at the moment, and what areas to focus on next.…”