All SoTA neural models underperform BM25 on Touché 2020, but why? 🤔 Our #SIGIR2024 paper re-evaluates neural models on Touché 2020, the argument retrieval task in BEIR (& MTEB)! 🍻 ❌ A significant issue is the bias towards retrieving short, non-argumentative passages! 1/8 🧵

.Maik Fröbe now presenting Resources for Combining Teaching and Research in Information Retrieval Coursework at #sigir2024

And next at #SIGIR2024 we have @nandan__thakur presenting work on re-evaluating neural retrieval models on Touchè 2020.

.Niklas Deckers now presenting "Evaluating Generative Ad Hoc Information Retrieval" at #sigir2024

ReNeuIR at #SIGIR2024: 61 systems evaluated within our shared task initiative. Submissions are still open!

Goodbye Washington! We had a fantastic week with interesting talks, discussions, and new ideas at #SIGIR24 #SIGIR2024. We hope to see you all again next year in Italy :)

Jan Heinrich Merker presented our short paper at #ArgMining2024 at #ACL2024 today. We proposed to add "semantics" to lexical retrieval models for argument retrieval. w/ Matthias Hagen, Maik Fröbe, Danik Hollatz. Paper: aclanthology.org/2024.argmining…

Great presentation about RAG in biomedical domain Jan Heinrich Merker #CLEF2024 #BIOASQ2024

Jan Heinrich Merker talks about how web search results of different quality influence decision-making #CLEF2024

IR legend Sean MacAvaney on stage at #haystackconf on re-thinking re-ranking. Representing Glasgow IR Group

Great read! Also a relatively contained example github.com/huggingface/sm…

Il y a un agresseur : la Russie. Il y a un peuple agressé : l'Ukraine. Nous avons tous eu raison d'aider l'Ukraine et de sanctionner la Russie il y a trois ans et de continuer à le faire. Nous, c’est les Américains, les Européens, Canadiens, Japonais et plusieurs autres.

We just released "German Commons", the largest openly-licensed German text dataset for LLM training: 154B tokens with clear usage rights for research and commercial use. huggingface.co/datasets/coral…