
happy to share our📜A Lightweight Constrained Generation Alternative for Query-focused Summarization - has been accepted to @SIGIR2023 as a short paper. work w/ @dcohenIR and more details 👇🧵
Daniel Cohen
29 posts


happy to share our📜A Lightweight Constrained Generation Alternative for Query-focused Summarization - has been accepted to @SIGIR2023 as a short paper. work w/ @dcohenIR and more details 👇🧵

Draft made available on arXiv ! 👇arxiv.org/pdf/2304.11721… We picture this constrained generation approach as a way to address the classical snippet problem in web search 🔍 and to help reduce users' cognitive overhead in search sessions 📃.





Excited to share that our full paper "Not All Relevance Scores are Equal: Efficient Uncertainty and Calibration Modeling for Deep Retrieval Models" was accepted at #sigir2021! Coauthors: @UnderdogGeek Oleg Lesota @navidrekabsaz @CarstenEickhoff. Stay tuned for the preprint!















Microsoft researchers have been exploring adversarial and reinforcement learning-based approaches to information retrieval. See what they're presenting at #SIGIR2018 about the recent work: aka.ms/AA1symr

Open Science via @mdr staff.fnwi.uva.nl/m.derijke/open…




Happy to announce: we'll organize a workshop on "Learning from Limited / Noisy Data for Information Retrieval" @sigir2018 #SIGIR2018 w/ @m__dehghani @diazf_acm @dr_hang_li @nick_craswell