Doctoral students

Ludovica Pannitto: ANNs – LSTMs in particular – have shown impressive results on tracking what seems to be abstract grammatical knowledge, without any explicit encoding of hierarchical structure. Ludovica investigates whether this empirical success can help us understand which kind of grammatical knowledge is learnable from a linguistic input when no prior innate structure is at hand. Web

Alexandre Kabbach: Accounting for the variability of word meaning across speakers of the same linguistic community is a fundamental challenge for any semantic theory. Alex investigates the social nature of lexical acquisition, asking whether conceptual similarity across speakers is a necessary requirement for successful communication. (Co-tutelle with the University of Geneva, supervisor: Prof. Jacques Moeschler.) Web

Masters students

2019 - 2020

  • Darya Andreyeva Darya is working on adapting the notion of negative sampling to the task of reference. Her system learns to name objects in real-world images from scratch.

  • Greta Gandolfi Greta is studying how biases

  • Siavosh Sepanta Siavosh studies the impact of different data sources on the ability of a distributional system to acquire core semantic competences.

2018 - 2019

  • Andrea Bruera: Andrea worked on modelling proper names in distributional semantics, showing that names cannot be dealt with using the kind of standard algorithm used for common nouns. The associated paper is on its way!

  • Elizaveta Kuzmenko: Liza took on the task of comparing what we see in the world and what we say about it. The first part of her work was published as a paper at the IWCS 2019.

  • Gosse Minnema: Gosse investigated the properties of the distributional spaces obtained via decoding of fMRI brain images. His analysis was published as a paper at the Student Session of ACL 2019. He also worked on the representation of named events in vector spaces.

  • Simon Preissner: Simon took the olfactory neural circuitry of a fruit fly and transformed it in a distributional model! He got best student paper award for his work at Clic-It 2019.

2017 - 2018

  • Stefano Bellelli: Stefano worked on learning word vectors from tiny data (one sentence only!) He captured ngram patterns correlated with the informativeness of a sentence.

  • Josine Rawee: Josine wrote her thesis on the semantics of color terms. She showed discrepancies between the ‘natural’ color of things in the real world and the color terms English uses when describing particular categories.