Fake news detection team

From urban legend to fake news. A global detector of contemporary falsehoods

Project: Infostrateg (NCBR)

The aim of the project is to make a large-scale computer-assisted content analysis of Polish writing, with particular emphasis on Internet content. The analysis will concern the occurrence of Fake News (hereafter: FN) and sensationalist motifs in the surveyed corpus of texts, and will result in the creation of a catalog and documentation that will serve as tools for the study of those aspects of the Polish Infosphere that relate to circulating narratives of a sensationalist (gossip, rumor, legend) or intentionally deceitful nature. The perception and interpretation of the world through sensational categories (e.g., conspiracy, anomaly, secret connection, omen/miraculous sign, action of “aliens” or supernatural forces, plague, decline of customs) is a significant and poorly recognized element of traditional thinking, which has been present in our culture for centuries and plays an important and constant role in many areas of social life. The study of this strand of media (information) will therefore at the same time contribute to a better understanding of some of the phenomena of modernity.

It’s easier to fool people, than it is to convince them that they have been fooled.

Mark Twain

The project will result in a system (DETECTOR) that will provide a modern tool for further empirical research into the significance of FN. Both the catalog and the computer software produced under the project will significantly enrich the set of research tools of various disciplines in the humanities. The project is implemented with the use of the latest (state-of-the-art) machine learning methods, such as neural networks, Natural Language Inference techniques, REALM-type models, etc.

Team leader

prof

Piotr Wierzchoń

Research: chronologization of Polish vocabulary of the 19th and 20th centuries, historical lexicography, automation of natural language processing, translational lexicography, disinformation theory, photo-documentation, machine learning: text classifiers, image recognition, time series analysis.

Team members

M.Sc.

Łukasz Borchmann

A specialist in natural language processing and computational linguistics; outside of the CSI team he is affiliated with Applica.ai, where he works as a Senior Research Scientist.

M.Sc.

Piotr Jabłoński

Doctoral student, conducting research on an automated model for comparing content in news media. The ongoing work aims to classify and detect misinformation in media messages.