BERT is the word

Within Social Science and Humanities (SSH), computational techniques such as NLP have been increasingly used. In this workshop, we will try our hand at one of the most precise and State-of-the-art models - BERT, which is pre-trained from unlabeled data extracted from a very large book corpus and most of the English Wikipedia.

There will be a brief introduction to BERT, however the primary focus of this workshop will be about trying the model yourself, through Google Colab python scripts, inspired by material from Bert for humanities

We will use data obtained from SCOPUS to investigate word similarities in abstracts from academic articles. To look more closely at how words can change meaning over time, how a term can be used in very different contexts, and therefore also have many different meanings.