Without going back to the recent "diplomatic incident" of a journalist for a major american daily newspaper that accused Emmanuel Macron file muslim school children, we must admit one thing : the fake news still has beautiful days in front of them. While scientists and engineers work with rigour, but under a certain pressure of public opinion, to build algorithms capable of detecting fake news, we watch with despair at the spreading of false truths by the masters of the new. It is enough to lower the arm when it does not scalp the hair. Under this prism dystopian in which the algorithms are improving the face of the individuals who place them, the fake news have an assured future !
It was thought the fake news politico-socio-sanitary behind us with the recent results of the american election and the announcement of two vaccines against the Covid-19. The disappointment is big and stuffy. It is pointed out that the fake news is part of a broader set of tools of misinformation that also contain the rumors or spam. These fake news are difficult to detect automatically and without error. That being said, the algorithms allow, by statistical analysis, by comparison with fake news earlier, or by the contribution of the humans themselves, to emphasize good number of them. And that starts with the analysis of their morphology.
A solution algorithms for the detection ?
The fake news have propagation velocities by the number of shares, likes and comments of their own. This allows algorithmically, similarity statistics, with the fake news already identified, to alert them to their possible existence. The analysis of the profile of users who share systematically the fake news also provides some interesting insights. Not to mention the tone of the text associated with the false news and comments, as well as their evolution in time, which is also a sensor effective. Finally, the human contribution in the search for a consensus on the veracity of the new is a fundamental contribution and complementary to the actions of algorithmic. Individually and collectively, by alerting on a possible risk or validating the truth of a subject. One can well understand why the coaster human is a threat to the effectiveness of the algorithms. Even more when it comes professionals are expected to master the famous fast checking and investigation.
In this on-going effort of research teams, a laboratory of the university of Arizona, is trying to build a library, and its associated computing code, of fake and real news, called FakeNewsNet. A repository open, for the contribution of all of us humans, if autoalimenterait and would grow, in theory, exponentially. All in all, the false truth may be days darker ahead... but this will only be achieved with the help of responsible individuals and informed, there will not escape !
