I am Associate Professor at Università di Bologna since 2022 (after three years as Senior assistant professor; RTDb, in the Italian system). I have been working on the (cross-language) assessment of text looking at different aspects such as originality (e.g., plagiarism detection) and intent (e.g., propaganda, hate speech).
Visit my UniBO website for university matters.
Before landing at UniBO , I spent 5 years as a Scientist at the Languuage Technologies group of QCRI and 2 as research fellow at TALP .
PhD in Computing Science, 2012
Universitat Politècnica de València, Spain
MSc in Artificial Intelligence, 2009
Universitat Politècnica de València, Spain
MSc in Computing, 2007
Universidad Nacional Autónoma de México, Mexico
BEng in Computing, 2002
Universidad Nacional Autónoma de México, Mexico
Research includes:
Research includes:
Research includes:
Università di Bologna
91258 -
Natural Language Processing
- 5 cfu
99813 -
Advanced Research Skills Lab
- 3 cfu
B3520 -
Profession-based Research
- 8 cfu
B0385 -
Natural Language Processing
- 6 cfu
99797 -
Advanced Professional Skills Lab
- 3 cfu
B2696 -
Language, Technology, Research I: Computational Thinking
- 6 cfu
PhD Python Gentle Introduction
91258 -
Natural Language Processing
- 5 cfu
B0385 -
Natural Language Processing
- 6 cfu
96829 -
Profession-based Research
- 8 cfu
96949 - Advanced Skills CL2 - 3 cfu
33409 -
Into Language Technologies
PhD Python Gentle Introduction
92586 -
Computational Linguistics
- 6 cfu
96829 -
Profession-based Research
- 8 cfu
84929 -
Software Localization
- 5 cfu
92586 - Computational Linguistics - 6 cfu
84929 - Software Localization - 5 cfu
92586 - Computational Linguistics - 6 cfu
84929 - Software Localization - 5 cfu
Under construction
The five editions of the CheckThat! lab so far have focused on the main tasks of the information verification pipeline: check-worthiness, evidence retrieval and pairing, and verification. The 2023 edition of the lab zooms into some of the problems and—for the first time—it offers five tasks in seven languages (Arabic, Dutch, English, German, Italian, Spanish, and Turkish): Task 1 asks to determine whether an item, text or a text plus an image, is check-worthy; Task 2 requires to assess whether a text snippet is subjective or not; Task 3 looks for estimating the political bias of a document or a news outlet; Task 4 requires to determine the level of factuality of a document or a news outlet; and Task 5 is about identifying authorities that should be trusted to verify a contended claim.
Some human preferences are universal. The odor of vanilla is perceived as pleasant all around the world. We expect neural models trained on human texts to exhibit these kind of preferences, i.e. biases, but we show that this is not always the case. We explore 16 static and contextual embedding models in 9 languages and, when possible, compare them under similar training conditions. We introduce and release CA-WEAT, multilingual cultural aware tests to quantify biases, and compare them to previous English-centric tests. Our experiments confirm that monolingual static embeddings do exhibit human biases, but values differ across languages, being far from universal. Biases are less evident in contextual models, to the point that the original human association might be reversed. Multilinguality proves to be another variable that attenuates and even reverses the effect of the bias, specially in contextual multilingual models. In order to explain this variance among models and languages, we examine the effect of asymmetries in the training corpus, departures from isomorphism in multilingual embedding spaces and discrepancies in the testing measures between languages.
Under construction
Under construction