Joining the merry #sigir2024 -rejection gang: Happy to share our paper on the Set-Encoder is now on arXiv arxiv.org/abs/2404.06912. It features permutation-invariant listwise re-ranking with an ELECTRA-base model that is as effective as RankGPT4. Code and data available 😉

Rather than continuing with the ongoing quest to study where and how AI can replace humans, a group of researchers suggest to examine forms of human–AI collaboration. Check it out here: bit.ly/3IOvlBE

#TechNews #LLMs #AI #research

Now we have Ferdinand Schlatt giving a talk about investigating sparse attention on cross-encoders at #ecir2024

.Ferdinand Schlatt is giving a talk on 'Investigating the Effects of Sparse Attention on Cross-Encoders', including a nice advertisement for ReNeuIR Workshop @ SIGIR 2024

'Who Determines What Is Relevant? Humans or AI? Why Not Both?,' by Guglielmo Faggioli, @LauraDietz99, @CLAClarke, Gianluca Demartini, @Matthias_Hagen, Claudia Hauff 🇪🇺 🇺🇦 🇩🇪 🇳🇱, Noriko Kando, @EKanou, @MartinPotthast, Benno Stein, & Henning Wachsmuth, on relevance decisions by #LLMs . bit.ly/43pYa0G

We just finished an #OWSome hackathon at TU Dresden to prepare the Workshop on Open Web Search #WOWS at #ECIR2024 in Glasgow. In the last week, we brainstormed, implemented, and evaluated a diverse set of retrieval components. This was so much fun! opensearchfoundation.org/wows2024

How will conversational search AI pay for itself?
It may be native ads or product placement in generated answers. At #CHIIR2024 next week, we'll present a user study showing that many people don't recognize ads inserted by LLMs in generated search results: webis.de/publications.h…

How good are language models in simulating human follow-up questions? (and how to even assess this?)

'Simulating Follow-up Questions in Conversational Search' with Marcel Gohsen Nailia Mirzakhmedova Matthias Hagen Benno Stein Webis Group got accepted at ECIR2024 #ecir2024

Happy to share our (Matthias Hagen Maik Fröbe) paper 'Investigating the Effects of Sparse Attention on Cross-Encoders' has been accepted at #ecir2024 . A substantial amount of the token interactions in cross-encoders can be removed without impacting effectiveness.

This work shows that sequence-to-sequence cross-encoders are susceptible to token injection from their own prompt, with concerning implications given recent work in decoder-based rankers. We also observe the transfer of these biases to bi-encoders. Pre-print is coming soon!

Happy to share that our paper 'Analyzing Adversarial Attacks on Sequence-to-Sequence Relevance Models' w/ Maik Fröbe Sean MacAvaney Martin Potthast Matthias Hagen has been accepted as a full paper ECIR2024 #ECIR2024

👍The #wows24 organisation team offers #mentoring for the practical part of #cfp submitting to early-stage #students such as master‘s and bachelor‘s students!
opensearchfoundation.org/wows2024/

Submitting period:
Call for papers: starts January 2024
Call for software components: open

We invite you to participate in the #ECIR2024 Workshop on Open Web Search. Let's discuss, develop, and promote an open web search ecosystem together! The workshop encourages submissions of scientific papers and implementations of retrieval components. opensearchfoundation.org/wows2024

#SaveTheDate + get your papers + software ready
#wows24 – 1st International Workshop on #OpenWebSearch
opensearchfoundation.org/wows2024
👉Phd + early-stage (Master/Bachelor) students
#ComputerScience #InformationScience NGI4eu #InformationRetrieval #NLP ECIR2024 SIGIR Students

📌 Meet Open Search Foundation @[email protected] partner Webis Group at #Digitalgipfel in Jena

@Matthias_Hagen, @MartinPotthast and Benno Stein from Webis Group will talk about #Biases and #GenerativeAI :
🌐 t1p.de/pvqat
📆 20 and 21 November

#OpenWebSearch #OpenWebSearch Community #DigitalGipfel

Leo Boytsov TIREx would be another option with a focus on TREC-style shared tasks: tira.io/tirex. You can submit runs, but the focus is more on software submissions. We have an open call for software submissions as part of an upcoming #ECIR2024 workshop opensearchfoundation.org/wows2024

Our short paper “Stance-Aware Re-Ranking for Non-factual Comparative Queries” with alexander bondarenko, Maik Fröbe, and Matthias Hagen got accepted at #ArgMininig 2023 ☺️

Takeaway: Improve nDCG by moving docs that take no stance down the result list.

Argument Mining #EMNLP #NLProc

It was a great Touché workshop at CLEF #CLEF2023
Thank you co-organizers and participants! w/ Simon Luck, Léo Hemamou, and Valentin Barrière

We are glad to share the recording of the invited talk by Nicola Ferro Nicola Ferro titled 'Comparing IR System Performance Through Explanatory Linear Models'. Thank you, Nicola, for many exciting insights and a fruitful discussion!
Video: youtube.com/watch?v=sWlEnh…

🧵We wrote a proposal to establish SIGIR findings. We will then work towards presenting it to the SIGIR steering committee. However, we need endorsements (and RTs! please). The link is here: us21.list-manage.com/survey?u=75a75…