Data Science @ Regensburg goes Leipzig: RAG and more

Dear all,

We are very happy to announce our next Data Science @ Regensburg Meetup. Given the success of last year's venture to Berlin we decided to have another road show this year. This time we will meet in Leipzig where our local host will be the webis group at ScaDS.ai, and the man who made this possible is the one and only Maik Fröbe.

The programme starts to take shape (and has already seen its first update :-) Lukas Bieberich (Data Scientist at jabmit) will give us some real-world insight into the use of Retrieval-augmented Generation (RAG) in industry.

We now also have Ines Zelch (PhD student Friedrich-Schiller-Universität Jena) confirmed as a second speaker.

Watch out for more details to be confirmed.

Looking forward to seeing you in Leipzig,
Udo

P.S.: If you cannot make it in person, then you may want to join us via Zoom (simply sign up and we will send out the access details closer to the date).

----------------------

Programme (more speakers and details to be added):

Speaker
Ines Zelch (PhD student Friedrich-Schiller-Universität Jena)

Title
Advertising in RAG Systems

Abstract
Conversational search engines such as YouChat and Microsoft Copilot use LLMs to generate responses to web search queries, which allows to include advertisements directly within the generated responses, instead of separately placing them next to a response. This is similar to native advertising and product placement, both of which are very effective forms of subtle and manipulative advertising. The talk provides insights into this new advertising scenario, describes the user awareness of this kind of advertising, and shows possible approaches for detecting and blocking such advertisements.

Speaker
Lukas Bieberich (Data Scientist at jabmit)

Title
From Static Workflows to Autonomous Agents

Abstract
An introduction to agent-based architectures in the context of modern LLM applications. We’ll dive into the mechanics of tool use and clarify the distinct roles of inference engines, frameworks, and model capabilities. We’ll explore Retrieval-Augmented Generation (RAG), along with single- and multi-agent patterns, and compare them with workflow-based approaches. Using hands-on examples, we’ll highlight their respective strengths, limitations, and cost implications.