Speaker: Rafael-Michael Karampatsis
2nd Year PhD Student; CDT in Data Science
Title: The naturalness of software
Abstract: Natural languages (NLs) are characterized by repetitiveness and predictability. This fact has allowed successful modelling of NL with statistical approaches such as language models (LMs) for tasks like speech recognition and machine translation.
I will focus on the observation that source code is also natural due to the fact that it is created by humans.
Lastly, I will present our state of the art results on LMs for 4 programming languages and highlight the importance of these models for every day coding and software engineering tasks (e.g., autocomplete, bug fixing).
I would be glad to receive your feedback!