FNDeepML
Annotation Scheme
FNDeepML (Fake News Deep Markup Language) is a fine-grained annotation scheme created for taking a further step in the deception detection, particularly in fake news detection.
FNDeepML (Fake News Deep Markup Language) is a fine-grained annotation scheme created for taking a further step in the deception detection, particularly in fake news detection.
This annotation scheme address news annotation from two levels:
This annotation scheme address news annotation from two levels:
a) Structure: the first level analysed during the annotation process of news is the structure one which divides a news piece in different parts. This division is made by applying the inverted pyramid hypothesis, that provides to each part a different level of relevance.
a) Structure: the first level analysed during the annotation process of news is the structure one which divides a news piece in different parts. This division is made by applying the inverted pyramid hypothesis, that provides to each part a different level of relevance.
b) Content: the elements belonging to the content are annotated in the second level by following the 5W1H technique, a method used in journalism for answering the six key questions in a news story.
b) Content: the elements belonging to the content are annotated in the second level by following the 5W1H technique, a method used in journalism for answering the six key questions in a news story.
Full explanation of the scheme can be found in the Specification page.