ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Combat COVID-19 infodemic using explainable natural language processing models | JoVE Visualize

Home
Research Domains
Information And Computing Sciences

Combat COVID-19 infodemic using explainable natural language processing models

Jackie Ayoub¹, X Jessie Yang², Feng Zhou¹

¹Industrial and Manufacturing Systems Engineering, University of Michigan-Dearborn, 4901 Evergreen Road, Dearborn, MI 48128, United States of America.

Information Processing & Management|March 29, 2021

View abstract on PubMed

Summary

This study introduces an explainable AI model using DistilBERT and SHAP to detect COVID-19 misinformation on social media. The model enhances public trust in AI predictions, crucial for combating health misinformation.

Area of Science:

Artificial Intelligence
Natural Language Processing
Public Health

Background:

COVID-19 misinformation on social media poses significant risks.
Deep learning models like BERT show promise in misinformation detection.
Existing models often lack transparency, hindering public trust.

Purpose of the Study:

To develop an efficient and effective explainable AI model for detecting COVID-19 misinformation.
To enhance public trust in AI-driven misinformation detection through explainability.

Main Methods:

A dataset of COVID-19 claims was collected and augmented.
A DistilBERT model was trained for misinformation detection.
SHAP (Shapley Additive exPlanations) was employed for model explainability.

Main Results:

The DistilBERT model achieved high accuracy (0.972) and AUC (0.993) on the custom dataset.
The model performed well on a larger dataset (accuracy: 0.938, AUC: 0.985), outperforming traditional models.
Explainable AI (SHAP) significantly increased participant trust and willingness to share information.

Conclusions:

The proposed explainable AI model effectively detects COVID-19 misinformation.
Incorporating SHAP explanations improves user trust and engagement with AI-identified information.
This approach offers a promising strategy for combating health misinformation online.

Keywords:

Computer Vision And Multimedia Computation

Pattern Recognition

Combat Covid-19 Infodemic Using Explainable Natural Language Processing Models

Misinformation detection