Intensity-Based Sentiment and Topic Analysis. The Case of the 2020 Aegean Earthquake

After an earthquake, it is necessary to understand its impact provide relief and plan recovery. Social media (SM) crowdsourcing platforms have recently become valuable tools for quickly collecting large amounts of first-hand data after a disaster. Earthquake-related studies propose using mining natural language processing (NLP) damage detection emergency response assessment. Using tex-data provided by the Euro-Mediterranean Seismological Centre (EMSC) collected through LastQuake app Aegean Earthquake, we undertake sentiment topic analysis according intensities reported their users in Modified Mercalli Intensity (MMI) scale. There were 2,518 comments, reporting from I X being most frequent intensity III. We use supervised classification rule-set defined authors two-tailed Pearson correlation find statistical relationships between MMI users, polarities, topics addressed comments. The word among comments was: “Felt.” (SA) indicates that positive polarity prevails associated with lowest reported: (I-II), while negative higher (III–VIII X). identifies increase polarity. was intensity, followed seismic information, solidarity messages, response, unrelated topics, building damages, tsunami effects, preparedness, geotechnical effects. Intensities are significantly negatively correlated number Positive decreases soar demonstrated validity our first hypothesis, despite not finding Instead, could prove damage, lifelines affected, effects topis only highest MMI.