About
This prototype has been developed based on our research in the CyCAT project.
How are these aspects calculated?
-
Political bias of the publishers
The political bias of the publisher is gathered from an external source (Media Bias/Fact Check), which annotated the bias based on the publishers' i) political affiliations, ii) story choices (if they publish from both sides or just one), iii) use of biased wording to sway readers, and iv) rates of factual reporting. "Left" represents a liberal view and "right" represents a conservative view. -
Geographical location of the publishers
The publishers' location is determined by analysing the suffix of the URL (e.g., "bbc.co.uk" is based in the UK, "abc.net.au" is based in Australia). If this information is not available, we used 'whois' command to identify the country where the domain is registered. -
Geographical focus of the articles
We used a named entity recogniser (spaCy, trained using the en_core_web_trf model) to identify country names in each article. The most frequent country is selected as the geographical focus of the article.
-
Entity focus of the articles
We used a named entity recogniser (spaCy, trained using the en_core_web_trf model) to identify entities discussed in each article. We removed "Covid" as it appears in all articles. The most frequent entity is then selected as the entity focus of the article.