The advent of the Web and Online Social Networks (OSN) created an unprecedented amount of accessible information. These tools have allowed everyone to become a news medium by setting up a website, a blog, or simply creating an account on an OSN, thus enlarging the offer with a significant number of individual uncontrolled contributions. While promoting more democratic access to information, direct and unfiltered communication channels may expose readers to manipulative, biased, and disinformative content.
The pollution of the societal debate, the distortion of the political agenda, and other effects of misuse of the Web and OSN represent significant issues for modern democracies. Since no recognizable authority handles the responsibility for the quality of the shared information, every user acquires a “reputability” just by her activity on the Web. Moreover, recommendation and advertising systems embedded into online news platforms and OSNs, generate the so-called algorithmic bias facilitating the formation of alternative realities, feeding into the confirmation bias, and amplifying polarization. This makes it ultimately challenging to evaluate the “quality” of the accessed information.
Objectives
Our research aims at three different and complementary goals:
define a methodological, data-driven framework for the study of online debates built on top of AI/ML data enrichment pipelines and high-order, feature-rich, and dynamic network modeling;
tune and apply such a framework in the context of PIEs (Polluted Information Environments) to study how they form, evolve and to what extent their evolution can be predicted;
propose an inferential methodology to measure pollution (while assessing its statistical significance), evaluate the outcome of mitigation strategies, and estimate pollution evolution predictability.