CEDRR hands-on seminar: Computational approaches to semantic change
CEDRR invites you to a hands-on seminar with Dr. Vojtěch Kaše from the Computing Culture & Society Laboratory (CCS-Lab, University of West Bohemia).
This seminar introduces a computational framework to analyze semantic trends in diachronic textual corpora. This framework draws on the so-called distributional hypothesis of meaning, which posits that there is a correlation between distributional similarity and semantic similarity of words: words appearing in similar language contexts tend to have a similar meaning. It implies that it should be possible to use the former (i.e., data on the distribution of words within a language corpus) to approximate the latter. Further, following that logic, when analyzing temporal trends in diachronic corpora, we can consider the option that a change in distribution implies a change in meaning. Formally, this approach relies to a substantial extent on the toolbox of linear algebra, representing distribution of words by means of multidimensional vectors, the so-called word embeddings.
In the hands-on part of the workshop, participants will “play” with some building blocks of this approach, using pre-prepared scripts in Python. The participants of the seminar will also explore the potential of treating words as multidimensional vectors and learn how to employ them in the case of diachronic analysis. In the afternoon session, dr. Kaše will introduce some examples of this approach from his own research.
Room: B2.12 (also known as B2.113)
Timeline: 9.12.2022, 9:30-12:30 - hands-on; 14:00-15:00 - examples
The event will be held physically but in case you are not able to participate directly, the lecture will be streamed on this Zoom link.