Detecting COVID-19 Misinformation on Social Media

Project Title

Detecting COVID-19 Misinformation on Social Media

Project Area of Specialization Artificial IntelligenceProject Summary

 The worldwide spread of COVID-19 has provoked wide online discussions, creating an ‘infodemic’ on social media platforms such as Twitter and Facebook. However, since most of the content on social media is not authored by medical professionals an extensive amount of health-related information being posted there has very low credibility and largely inaccurate and misleading. Such health-related misinformation, thereby not only leads potentially ill people away from proper treatment and care and disrupting the lives of common people but also being used as a tool to disrupt the economy of countries, reduce people’s trust in their governments, and to promote different products for profitability.

To address this misinformation spread problem, we propose an approach to detect misleading information about COVID19 from social media, such as tweets using NLP models.

Specifically, we intend to develop a web-based tool that employs machine learning and Natural Language Processing to detect whether a given Tweet text has a misconception, and if so, whether the discussion propagates or agrees with the misconception or disproves the misconception.

 High-level idea of the approach: We intend to employ a two-step approach for the misconception detection problem stated above:

1. Misconception Retrieval: Given a Tweet, identify a set of related misconceptions from a knowledgebase of known misconceptions.
2. Stance Detection: For each misconception-tweet pair, predict whether the misconception and tweet text Agree, Disagree, or tweet text takes No Stance with respect to misconception.

Project Objectives

We intend to develop a web-based tool to detect COIVID-19 related misinformation in social media posts. The broader objective is to prevent health misinformation diffusion in social media by flagging posts that spread misinformation.

Project Implementation Method

1. Tweet Collection (using Twitter API Tweepy).
2. Building a database of known misconceptions about COVID-19 pandemic and vaccine.
3. Misconception retrieval
4. Stance detection
5. Results evaluation
6. Web-based interface development

Benefits of the Project

1. Our project will help the society to differentiate between credible and misleading information related to COVID-19 and its vaccine.
2. It will help in detecting misleading information on any future global health issues, such as second and third wave of Corona Virus and vaccines.
3. Save people seeking health-related information on social media from fraudulent content and guide them towards proper treatment and care.

Technical Details of Final Deliverable

Technology Domain:  Machine Learning

Programming Language:  Python

Tools: Google Collaborator, Sublime Text 3, MySql

Final Deliverable of the Project Software SystemCore Industry HealthOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	39250
16GB RAM	Equipment	1	8000	8000
2TB Hard Disk(External for Tweets Collection	Equipment	1	12000	12000
Google Colab license (USD 9.99 x 4 months)	Equipment	3	6417	19250