Italian by origin, I've lived and studied in Trento and Groningen, got my PhD in Antwerp and spent a couple of months in Tübingen as a visiting research. Now I live and work in Tilburg in the Department of Cognitive Science and Artificial Intelligence.
I work on computational models of language learning, particularly investigating how the relation between how a word sounds and its meaning may help children learn language. With a study on this topic I won the best student presentation award at the 11th International Conference on the Mental Lexicon in 2018. I also do work on sequence modelling in more data scienc-y applications, such as user intent prediction from clickstream data. My interests also include cognitive science and multimodality.
Outside of the office I like cooking, playing volleyball, improvising on stage, cycling on the hills around my home town, enjoying a book, a concert, a movie or a play.
I've worked a lot with distributional semantic models: I've used these models to study how semantic representations developed by congenitally blind and sighted people differ (or not), how children combine linguistic and visual inputs to ground concept learning, and how word forms map onto their meaning representations. In doing this, I've used a variety of statistical and machine learning methods, including Generalized Linear Mixed Models, Generalized Additive Mixed Models, perceptron-line neural networks, LSTMs, Naive Bayes classification, Linear Discriminant Analysis, random forests and kNN algorithms. I mostly program in Python and R, but did some stuff in Matlab some time ago.
I published my research in journals such as PLoS One and the Journal of Experimental Psychology: Learning, Memory, and Cognition and presented my work at international and national conferences, including the Annual Meeting of the Cognitive Science Society.
The master course in Natural Language Processing for the DSS master focuses on core NLP problems and tools and how they are used to develop end-to-end applications such as language generation, text classification and machine translation are reviewed. Students learn to design an NLP tool to address a specific problem (choose the data, pre-process it, extract the relevant information, evaluate the system and improve on it). This course assumes a non-technical audience.
The bachelor course in Computational Linguistics for the CSAI program (3rd y) covers similar topics but focuses more on the algorithms and the solutions developed in the last decades, to provide students with the ability to implement these solutions. The relation between computational linguistics, cognitive science and AI is also investigated. The focus is more the applied and a rather technical audience is assumed.
I also teach the R practical sessions for the Statistics for CSAI II (CSAI bachelor, 2nd y)
If you work or wish to work on the relation between form and meaning from an experimental or modelling perspective, drop me a line. I'm also interested in extending collaborations on more applied tracks, especially sequence modelling in social science and education contexts.
At the moment, I collaborate with Tooso, a start-up specialised in developing ai and NLP solutions for e-commerce websites. So far, we have investigated the problem of user intent prediction using click-stream data, i.e. determining whether a user is going to buy something or not based on his or her click on a website. I also collaborate with Marco Marelli, from the University of Milano Bicocca, on research about the relation between form and meaning and its influence on how children learn language, people process it and languages evolve to be learned and spoken.