On the theoretical side of computational linguistics, I am interested in categorical and logical aspects of natural language and, in particular, in reconciling symbolic and statistical approaches.
I investigate the mathematical structure of contextuality and ambiguity in various models of linguistic meaning. Ultimately, I wish to clarify how meaning emerges out of composing discourse fragments (from words to sentences, sentences to text) and how we resolve ambiguity by refering to a broader context (from text to sentences, sentences to words). To tackle these problems I use the tools of category theory, logic and machine learning.
More recently, I have become interested in the structural properties of sequence to sequence models in NLP, especially in relation to automata and formal languages, their discrete counterpart.
After a degree at HEC Paris I studied mathematics at Université Pierre et Marie Curie (Paris VI) and completed a masters degree in computer science at the University of Oxford. My MSc. thesis explored abstract and concrete models of lexical ambiguity in compositional distributional semantics.
Open System Categorical Quantum Semantics in Natural Language Processing
Robin Piedeleu‚ Dimitri Kartsaklis‚ Bob Coecke and Mehrnoosh Sadrzadeh
In CoRR. Vol. abs/1502.00831. 2015.