The Center for Positive/Empirical Analysis of Political Economy and the Graduate School of Political Science, Waseda University will hold a special seminar titled “A practical introduction to word embedding techniques for social science – with an extension to LLM” by Prof. Luigi Curini on January 29, 2024. We would like to introduce this opportunity and look forward to your participation.
Date & Time: Jan. 29 (Fri) 13:10-16:45
Venue: Waseda Campus, Bldg. No. 3, 9F Room #902
Registration is not necessary.
Language：English
Organized by: Graduate School of Political Science / Center for Positive/Empirical Analysis of Political Economy
Contact：Airo Hino [email protected]
Please see below for the abstract from Prof. Curini:
This seminar is divided in two parts. In the first one, I will give an introduction to word embedding (WE) techniques, focusing in particular on GloVe (Global VEctors of words) model. The aim is a) to present the main ratio behind word embedding models; b) to explain why social scientists should care about them (as a method on their own for analyzing text as data, but also as useful complements to other algorithms, such as machine learning algorithms, dictionaries, and so forth); and c) to give some advice to practitioners. I will also briefly discuss about recent advancements in the literature concerning LLM and contextual embeddings (with specific attention to BERT). The second part of the seminar will offer a hands-on lab on how to fit a GloVe model in R, as well as BERT. A general knowledge of R (in particular of the Quanteda package) would be useful, as well as familiarity with machine learning literature. The scripts and the dataset to replicate the Lab-class will be made available the day before the seminar at the following URL: http://www.luigicurini.com/applied-scaling–classification-techniques-in-political-science4.html
