Building Swedish language models

Long time since the last meetup, but let's try with an online meetup and please try to be as interactive ask possible and ask lots of questions. We are going to try the BigMarker tool for this webinar, which has some nice interactive features and without any software installation requirements.

Note: meetup.com does not seem to support BigMarker, please go to the following URL and register for the event. https://www.bigmarker.com/peltarion/SWEBERT

The progress in NLP in recent years has been staggering. However, most of the research focuses on non-Swedish text, and we need to make sure that these techniques also work for Swedish use cases. This is critical, for example, when developing innovative and cost-efficient public services.

One of the most prominent Swedish initiatives in this direction is the Swedish language models for Swedish authorities project. The initiative is a three-year project funded by Vinnova, where RISE, LTU, Peltarion, the Swedish Public Employment Service, and more are participating in building up Swedish language models using the latest NLP techniques. We are just in the initial stages of the project, but we already have some initial results that we would like to share with you.

Agenda

• 18:00 Welcome and panel introduction, hosted by Anders Arpteg
• 18:15 Introduction by Vinnova, Daniel Rencrantz
• 18:45 Why is this important to the National Library of Sweden, Love Börjesson
• 19:15 Predicting political parties with multilingual NLP, Peltarion, Sebastian Callh and John Moberg
• 19:45 Final discussions and Q&A
• 20:00 The end

It is free for anyone to join. We will open the webinar at 17:45. Please join early to have time to register and get the video conferencing working properly. The webinar will be live-streamed and recorded.