Data-driven analytics of COVID-19 'infodemic'

Area of Science:

• Computational Linguistics
• Public Health Communication
• Psycholinguistics

Background:

• The COVID-19 pandemic was accompanied by a significant infodemic, challenging information credibility.
• Existing research on misinformation often overlooks linguistic nuances and consumer psycho-social behavior.
• Current fact-checking and labeling models lack a deep understanding of language characteristics.

Purpose of the Study:

• To identify lexical and grammatical features of COVID-19 misinformation.
• To analyze psycho-linguistic triggers (sentiment, power, activity) using Affective Control Theory.
• To develop feature indexing for anti-infodemic modeling.

Main Methods:

• Data-driven analysis of COVID-19 misinformation content.
• Linguistic feature extraction, focusing on sentiment, power, and activity.
• Application of Affective Control Theory for psycho-linguistic interpretation.
• Development of feature indexing for misinformation detection models.

Main Results:

• Misinformation exhibits distinct language patterns, favoring evaluative terms and multimedia.
• Negative sentiment is a prominent feature used to engage audiences.
• Appeals to sympathy and emotional triggers effectively encourage information sharing.

Conclusions:

• COVID-19 misinformation employs specific linguistic strategies to evoke emotional responses and promote spread.
• Understanding these psycho-linguistic triggers is crucial for effective anti-infodemic interventions.
• Feature indexing based on these linguistic patterns can enhance misinformation modeling.