how to create your own web page

Propaganda
Analysis Project

Developing tools to help readers detect the use of propaganda and analysing how propaganda and extremely-biased articles spread online.

Latest News and Activities

  • Accepted Poster @AI for Social Good Workshop at NEURIPS'19 - Giovanni and Seunghak will present the work on fine-grained propaganda detection for the first time at a machine learning event

Invited Talk @Facebook London - Giovanni presented Proppy and the work on fine-grained propaganda detection. Slides available. (Oct. 2019)

Accepted Poster @Truth and Trust Online Conference 2019 - Giovanni and Preslav presented the work on fine-grained propaganda detection. Poster available. (Oct. 2019)

Proppy at "Notte dei Ricercatori" - Notte dei ricercatori (night of researchers) is an annual event in which researchers organise meetings and demonstrations to explain their research to the general public. Alberto showcased Proppy.
Slides in Italian available. (Sep. 2019)

RESEARCH

Mobirise

Persuasion Techniques Detection

Propagandist messages are conveyed through a number of rethorical and psychological techniques. 

Papers and Activities

We annotated news articles at the fragment level with 18 persuasion techniques and developed our own machine learning model to automatically detect them [3]; the code and the data of our system are available. You can try a demo of our model here.

State of the Art - Test Your Own Models

We are currently organising a shared task at SemEval 2020. After the task is over, we will keep a live leaderboard for the users to compare the performances of their systems and to keep track of the state of the art. 

Mobirise

Proppy: Document-level Propaganda Detection

Proppy is a system that organizes news articles into events and, for each event, shows articles according to their level of propagandistic content. We aim at raising awareness into individual readers as well as providing tools for organizations to monitor large amounts of news articles.

Papers and Activities

We analyse various representations, from writing style and readability level to the presence of certain keywords for detecting propagandist documents [2]A demo of our work continuously collects articles, organises them into events and their propagandist content, as determined by Proppy [1]

CORPORA

Mobirise

Persuasion Techniques Annotation

Among all persuasion techniques defined in the literature, we have selected the ones that can be found in news articles and the ones that can be spotted intrinsically, i.e., just by reading the article, without resorting to external knowledge. The final list consists of 18 techniques.
Currently, we have annotated 500 articles in collaboration with A Data Pro. The annotation process is described here

Mobirise

Propagandist Documents Collection

For our experiments with Proppy, we collected 51k propagandist and non-propagandist articles, labelled using distant supervision according to the labels of the sources provided by Media Bias/Fact Check. The corpus is available for download here

EVENTS

Sep. 2019 - Jan. 2020

SemEval 2020 Task 11 
"Detection of Propaganda Techniques in News Articles"

As a further effort to develop/foster a community around the topic of propaganda detection, we are organising a shared task at SemEval 2020.

3 Nov. 2020

Tutorial on "Fact-Checking, Fake News, Propaganda, and Media Bias: Truth Seeking in the Post-Truth Era" @EMNLP'2020

As a further effort to develop/foster a community around the topic of propaganda detection, we are organising a shared task at SemEval 2020

Past Events

  • Workshop on "Bias, Disinformation, Misinformation and Propaganda in Online News and Social Media" @ Socinfo'2019 - We co-organise the workshop, which will be held in Doha, Qatar. A sizeable number of the submissions were propaganda-related. (Nov. 2019) 
  • NLP for Internet Freedom Workshop (NLP4IF) Workshop @EMNLP'19 - The workshop included a shared task on detecting propaganda techniques. (Nov. 2019) 
  • Hack The News Datathon - we co-organised a Hackathon on propaganda detection at the document-, sentence- and fragment-level (Feb. 2019).

PUBLICATIONS

3

G. Da San Martino, S. Yu, A. Barrón-Cedeño, R. Petrov, P. Nakov, "Fine-Grained Analysis of Propaganda in News Articles", in Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019), Hong Kong, China, November 3-7, 2019. 
Bibtex - Code and Data

2

A. Barrón-Cedeño, I. Jaradat, G. Da San Martino, P. Nakov, “Proppy: Organizing News Coverage on the Basis of Their Propagandistic Content”, Information Processing and Management, 2019. DOI:10.1016/j.ipm.2019.03.005. 
Bibtex

1

A. Barrón-Cedeño, G. Da San Martino, I. Jaradat, and P. Nakov, “Proppy: A System to Unmask Propaganda in Online News”, in Proceedings of the Thirty-Third AAAI Conference on Artificial Intelligence (AAAI’19), Honolulu, HI, USA, 2019.
Bibtex - Online Demo

TEAM 

Giovanni
Da San Martino

Scientist
QCRI, HBKU

Alberto
Barrón-Cedeño

Research Assistant
Università di Bologna

Preslav
Nakov

Principal Scientist
QCRI, HBKU

Seunghak
Yu

Joint Post-doc
QCRI - MIT CSAIL

Alumni

Israa 
Jaradat

PhD Student
University of Texas at Arlington

Email: propaganda_analysis_project@googlegroups.com         Twitter: _proppy_