Skip to main content
  • Expression of Interest

    NLP and IR Group, UNED

    Related to the call, our group is currently working on detecting misinformation in tasks such as fake news and stance detection. We work under the hypothesis that besides The content of the message (text or meme), it is important to determine the source and the intention of the communication, and profiling these sources. For this purpose, we are analysing not only the text, but also users’ profiles and their social network connections. We have already found that the combination of several sources improves the detection of misinformation and we continue researching the best way of combining these sources.

    We have tested our hypothesis in several datasets. Related to stance detection, we have worked on Brexit and SemEval-2016 task 6. Related to fake news detection, we have used FakeNewsNet. Besides, we have participated in the SardiStance 2020 competition at Evalita, where we obtained the best results combining several sources of information for detecting stance in Tweets [1].

    We have organized the VaxxStance 2020 evaluation campaign at IberLef [2,3], aiming at the benchmarking and evaluation of stance detection systems in the domain of “vaccines” over twitter in several languages. We collected and annotated tweets related to vaccines together with their context information in twitter to create a gold standard that enabled systems evaluation.


    [1] Rodrigo Agerri, Roberto Centeno, María Espinosa, Joseba Fernández de Landa, Álvaro Rodrigo. VaxxStance@IberLEF 2021: Overview of the Task on Going Beyond Text in Cross-Lingual Stance Detection. Procesamiento del Lenguaje Natural, Revista nº 67, 2021.

    [2] Maria S. Espinosa, Rodrigo Agerri, Alvaro Rodrigo, Roberto Centeno. DeepReading@SardiStance 2020: Combining Textual, Social and Emotional Features. Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2020).

    [3] Julio Gonzalo, Manuel Montes-y-Gómez, Paolo Rosso. IberLEF 2021 Overview: Natural Language Processing for Iberian Languages. Proceedings of the Iberian Languages Evaluation Forum (IberLEF) 2021.

    The UNED IR & NLP Research Group ( started its activities in 1996 and now hosts over 20 researchers working in several NLP & IR research topics such as Fundamental Research in NLP & IR, Online Reputation Management, Text Mining in Social Media, Text Mining in BioMedical Domains, Open Information Extraction and Knowledge Capture and Question Answering. Their researchers has led several research projects, both national and international, related to NLP. The group has also a vast experience organizing evaluation campaigns at different forums such as CLEF, SemEval, IberLef, etc. Their members published their  research in the most important conferences, e.g. ACL, SIGIR, EMNLP, etc, and relevant journals.

    The team for the current CHIST-ERA call is composed of the following three members:

    • Anselmo Peñas, Full Professor, full time . He has participated in numerous national and European research projects in the fields of Artificial Intelligence, Access to Information and Natural Language Processing, fields in which he has published more than 80 articles. From 2007 to 2015, he coordinated the European multi-language question-answer systems evaluation campaigns (CLEF QA). He has also been international coordinator of the CHIST-ERA READERS project (2013-2015) and principal investigator of the CHIST-ERA LIHLITH (2018-2020).
    • Roberto Centeno, Professor, full time His main research interests are focus on areas of social media, fake news and stance detection on social networks, reputation & trust mechanisms and regulation of open systems. He has authored several research papers in journals and specialists workshops in the field of artificial intelligence and multi-agent systems. He has participated in several research projects focus on the application of Artificial Intelligence techniques to real world problems. He has taken part in the organization of the VaxxStance evaluation at IberLEF 2021.
    • Álvaro Rodrigo, Professor, full time He researches on Natural Language Processing, mainly in the areas of Question Answering systems and disinformation analysis. He has taken part in the organization of Question Answering evaluations at the Cross Language Evaluation Forum from 2006 and the VaxxStance evaluation at IberLEF 2021. He has also been involved in several Spanish research projects. He serves as reviewer for several international journals and conferences.