Francesco Barbieri

Research Intern - Snap Inc.

Phd Candidate - Universitat Pompeu Fabra (LaSTUS)
Mail: francesco.barbieri at upf dot edu
Office: 55.424

Google Scholar

I am in the fourth year of my Ph.D in NLP at UPF Barcelona and plan to defend my thesis in January 2018 (Horacio Saggion is my PhD advisor). I study figurative language that expresses irony, sarcasm and satire through Twitter posts. I also explore the visual aspects of social media, analyzing images posted online and multimodal forms of communication such as emojis.

I received my bachelor's degree in Computer Science from the University of Bologna in December 2010 with a final thesis on quantum computing (supervised by Ugo Dal Lago).

I then studied Cognitive Science at University College Dublin and devised a M.Sc. thesis (2013) on irony detection in Twitter data (supervised by Tony Veale).

From January to April 2016, I visited research scientist Marco Baroni's group at the University of Trento and worked on Deep Learning Multimodal systems.

Curently, I am working on multimodal systems at Snapchat (Aug'17 -> Dec'17).

I am also co-organizing a Semeval task on emoji prediction in English and Spanish. Registration is open!




Best paper award in WNUT-EMNLP September 2017 for the paper: Towards the Understanding of Gaming Audiences by Modeling Twitch Emotes.

Best oral presentation award ISMIR October 2017 for the paper: Multi-Label Music Genre Classification from Audio, Text and Images Using Deep Features

Best paper award in SEPLN September 2015 for the paper: Is This Tweet Satirical? A Computational Approach for Satire Detection in Spanish


Some Resources

Dataset on satire, irony and humor in Twitter from my publications (only IDs as we are not allowed to share tweets direcltly)

USA Twitter embeddings: the focus of the paper is on emojis, but these embeddings can be used also for words. Updated the 15th of February 2016

How Cosmopolitan Are Emojis?  We studied emojis in different languages, you can find our embeddings in the link.


Media Coverage


Crean inteligencia artificial predice los emoticonos más adecuados a un texto El Periodico

Ya es posible hacer una predicción de los Emoji que le corresponden a un texto El Economista

Emojis, ¿el nuevo lenguaje emocional ?- La Vanguardia Magazine

Tinder desvela los 20 emojis más usados por los españoles en su perfil - El Economista


Emojis, un nuevo lenguaje universal - La Vanguardia

Los emojis más utilizados en diferentes países - Las Provincias

¿Son los emoticonos universales? - Voz Populi

Les emoticones més utilitzades són un llenguatge universal - TV3

Crean un método automático que extrae significado de los Emoticonos teléfono - La Vanguardia

Un mètode automàtic extreu el significat de les Emoticones del telèfon - Diari de Girona


Todo al pie de la letra - La Vanguardia

Cómo distinguir un tuit satírico - Alef


Conferences and Summer Schools

I have published and participated to international conferences like ACMMM, IJCAI, EACL, RANLP, WASSA, LREC and national conferences such as SEPLN (Spanish) and CLICIT (Italian).

I also attended a few summershools, including the Lisbon Machine Learning Summer School (LxMLS) and the Computational Creativity Code Camp (ICCCC)