Data Exploration and Classification of News Article Reliability: Deep Learning Study.

Area of Science:

• Natural Language Processing (NLP)
• Machine Learning (ML)
• Deep Learning

Background:

• The COVID-19 pandemic created an "infodemic" characterized by the rapid spread of misinformation.
• Misinformation undermines public health policies and orders.
• Manual fact-checking is insufficient to combat the scale of online misinformation.

Discussion:

• This research introduces an ensemble deep learning model to classify news article reliability.
• The model utilizes text body, sentiment, lexical categories, and readability features.
• Analysis revealed reliable articles exhibit neutral sentiment and higher readability compared to unreliable ones.

Key Insights:

• Unreliable news articles often feature negative sentiment and lower readability.
• The developed model achieved an Area Under the Curve (AUC) of 0.906, with 0.835 specificity and 0.945 sensitivity.
• These performance metrics surpass the baseline of the original ReCOVery model.

Outlook:

• This work identifies novel linguistic markers distinguishing reliable from unreliable news.
• The findings can empower researchers and the public to identify false information.
• Future applications include real-time detection of misinformation in online news media.