Thousands of state-sponsored propaganda tweets about the Iran nuclear deal were analyzed using artificial intelligence by experts at the University of Portsmouth.
The findings show that natural language processing and machine learning tools can help analyze large-scale data and gain insights into Iran’s social media strategy.
Computing experts started with an initial dataset of over 9 million tweets from nearly 6,000 accounts in 67 languages that Twitter identified as being associated with the Iranian state.
They filtered this dataset by language, date and keyword to leave over 24,000 tweets and analyzed for sentiment and objectivity. Sentiment analysis is used to assess opinions on a topic and determine emotional tone.
Lead author Michael Burroughs, who worked on the study while at the University of Portsmouth, said: “Tweet objectivity scores were high, particularly those labeled as emotionally neutral.” This supports the claim that there is a possibility that Given that social media platforms open their verification programs to paying customers, the potential for using objective language to influence discourse on various topics is enormous. “
In this study, we explored the possibility of automatically labeling and detecting whether tweets are positive, neutral, or negative using five machine learning algorithms.
“An increasing number of people are getting the latest news through social media,” said Ella Haig, Ph.D., artificial intelligence leader in the University’s Department of Computing. It means that it is being used to spread narratives and propaganda that can give.” Misinformation is a growing concern, making it increasingly difficult for users to distinguish between false and true information.
“We need an approach that uses automated processes to support the detection and analysis of misinformation. phenomenon. “
