There is a lot of excitement and hype around ChatGPT and other LLMs, and they are making impressive progress. In this talk, I'll discuss how LLMs can be applied to web data extraction, what are the possible paradigms and their pros and cons. We'll cover code generation, selector generation, online data extraction, using open source models and commercial services, and check their cost and quality. I'll cover research and community projects, and will share our experience using ChatGPT and open source models for internal projects and prototypes.

Konstantin is the Head of Data Science at Zyte where he leads Machine Learning research and development. He has also participated in Kaggle competitions, achieving a Grandmaster Title, and contributes to the community with talks, sharing code, and knowledge.