Oliver Flasch's Blog

Risk Assessment in AI Projects

Thoughts on risk assessment in AI-, machine learning-, and data science-projects.

Topics in the Bundestag (Part 2)

In the first part of this series, we employed deep learning to classify all speech paragraphs of the current German parliament into 13 predefined topics. In this part, we will finally answer our initial question: Who talked about what (and when) in the 19th Bundestag?

NLP

Topics in the Bundestag (Part 1)

Who talked about what in the 19th German Bundestag? To answer this question, you'd probably need a panel of hardworking undergraduates reading and categorizing hundreds or even thousands of speeches. Maybe we can apply modern Natural Language Processing to automate this task at least partially?

Data Science

German Plenary Proceedings as an NLP Testbed

With the 19th electoral term of the German federal parliament drawing to a close, it could be interesting to do some data analysis on what has been said there. In this post, we will download the relevant plenary proceedings and assemble them into a clean dataset.

NLP

Deep Learning German

How to auto-complete your thoughts with a modern deep learning model, in German!

General

Read Me

This blog is about data strategy and artificial intelligence. I'm writing to capture ideas, walk through experiments and document what works in practice (and what doesn't). Here are some of