DisInfoNet is the new SOMA toolbox for data collection and analysis to combat the phenomenon of disinformation created by the Luiss Data Lab. DisInfoNet is a Python library including modules for managing archives, elaborating and classifying text, building and analyzing graphs, and more. DisInfoNet is also a toolkit with a pipeline designed for journalists and fact-checkers with no coding expertise. Through a user-friendly dashboard the members of the SOMA observatory will be able to assess the prevalence of disinformation in social media data.
DisInfoNet will allow investigating the correlation between users’ polarization and participation in disinformation campaigns, and highlighting the primary actors of disinformation production and propagation. DisInfoNet provides the ability to upload datasets taken from Twitter, in the JSON Tweet Format, and carry out analyzes on them. It is possible to create Filters to isolate, manage and analyze fake news or a set of related fake news through dynamic statistics and visualizations.
You can create one or more binary classifiers to cluster a dataset. Classifier is based on a semi-automatic “self-training” process, in which a list of hashtags associated with two classes of interest are used to automatically extract a training set. For each dataset, several interactive data visualizations are automatically generated, describing the quantitative, temporal and spatial distribution of the data.
The relevant Manual is available here