IRLab

Information Retrieval Lab @ University of A Coruña

[Workshop on Online Abuse and Harms @WOAHWorkshop]

🎉 Thank you to everyone who made #WOAH2026 possible!

A big thank-you to our speakers, authors, reviewers, sponsor, and participants for another day of discussion on online abuse and harms.

📍 See you at #WOAH2027 at ACL in Kyoto, Japan, in August!

More details coming soon.

[Workshop on Online Abuse and Harms @WOAHWorkshop]

🙏 A big thank-you to HYBRIDS for supporting #WOAH2026!

Your support helps bring the online abuse and harms research community together. Thank you for helping make this workshop possible!

#EMNLP2026

[paloma @itspalomap]

✈️ Heading to #EMNLP2026 in Budapest! 🇭🇺

📌 Presenting two papers on Tue, 27 Oct, 9:00–10:30:

🔹 From Self to Other: Evaluating Demographic Perspective-Taking in LLM Hate Speech Annotation

🔹 Beyond the Explicit: A Bilingual Dataset for Dehumanization Detection in Social Media

[paloma @itspalomap]

📌 Findings: Beyond the Explicit: A Bilingual Dataset for Dehumanization Detection in Social Media w/ @Dede198904 Katarina Laken @david__jurgens @clauwa @gesis_org

🔗 https://t.co/8mxHtdr2ZG

See you in Budapest! 🇭🇺

[paloma @itspalomap]

Excited to share that two of our papers have been accepted at #EMNLP2026! 🎉

📌 Main Conference: From Self to Other: Evaluating Demographic Perspective-Taking in LLM Hate Speech Annotation w/ @jparapar

🔗 https://t.co/mvtGr9tTs8

@IRLab_UDC @HYBRIDS_MSCA

[IRLab @IRLab_UDC]

See you in Budapest! 🇭🇺

#EMNLP2026 @emnlpmeeting

[paloma @itspalomap]

Excited to share that two of our papers have been accepted at #EMNLP2026! 🎉

📌 Main Conference: From Self to Other: Evaluating Demographic Perspective-Taking in LLM Hate Speech Annotation w/ @jparapar

🔗 https://t.co/mvtGr9tTs8

@IRLab_UDC @HYBRIDS_MSCA

[paloma @itspalomap]

This week I'm attending ICWSM 2026 in LA 🇺🇸☀️

I'll be presenting our paper: "Personalisation or Prejudice? Addressing Geographic Bias in Hate Speech Detection using Debias Tuning in Large Language Models" 🧠

🎤 Short oral: Wed 17:00-18:00
🖼 Poster: Thu 10:30-11:00

[Javier Parapar @jparapar]

Happy to share that the @IRLab_UDC team will present the following works at #SIGIR2026 🎉 Great work by everyone involved, very proud of these contributions! Abstracts below👇

[Javier Parapar @jparapar]

👉 In “Hybrid Pooling with LLMs via Relevance Context Learning” (with @davidoterof), we turn examples into explicit, topic-aware relevance narratives, rethinking how LLMs judge relevance. #SIGIR2026

[Javier Parapar @jparapar]

👉In “Beyond Top-e: Simulation-Based Interactive Evaluation for Query Suggestions” (with @JorgeGabin_ & @wangxieric ), we move from static assumptions to interactive simulation, using LLMs to model how users actually select query suggestions. #SIGIR2026

[Javier Parapar @jparapar]

👉In “Lost in the Evidence? Reproducing Document Position and Context Size Effects in RAG” (with @JorgeGabin_ & @anxopvila ), we revisit position and context effects in RAG, highlighting the role of evaluation design and reproducibility. #SIGIR2026

[Javier Parapar @jparapar]

👉In “Learning Evidence of Depression Symptoms via Prompt Induction”, (with Eliseo Bao, @anxopvila and @davidoterof ) we teach LLMs to detect symptoms by turning examples into clear, reusable guidelines. #SIGIR2026

[Javier Parapar @jparapar]

👉 Finally, in “Faithful LLM-Assisted Question Answering for Technical and Academic Search” @jlfilgueiras explores how to design LLM-based search systems that provide grounded, verifiable, and up-to-date answers. #SIGIR2026

[paloma @itspalomap]

Today we will be presenting our work "PartisanLens: A Multilingual Dataset of Hyperpartisan and Conspiratorial Immigration Narratives in European Media".

📍 Poster Board #41
🕓 16:30–18:00
📌 Poster Hall

🤝 If you're around, come say hi!

#EACL2026 #NLP

[Javier Parapar @jparapar]

Una reflexión personal: ¿y si la IA pudiera detectar a tiempo señales de malestar emocional en lo que escribimos? 👇

🔗https://t.co/muBL8blvRr

[paloma @itspalomap]

Paper accepted at #LREC2026 🎉

Mute Cods: A Multilingual Telegram Dataset with Benchmark Models for Conspiracy Theory Detection 🌐

w/ Katarina Laken, Erik Marino, Davide Bassi, Søren Fomsgaard, Michele Maggini, Renata Vieira, Marcos Garcia & Sara Tonelli

See you in Palma! ☀️📚

[Javier Parapar @jparapar]

🥳 @itspalomap paper "Can LLMs Evaluate What They Cannot Annotate? Revisiting LLM Reliability in Hate Speech Detection" has been accepted at #LREC2026 w/ @davidoterof & @patrimrodilla See you in Palma 🇪🇸✈️
🔗https://t.co/jxeH1xtbgj
📀https://t.co/zawV69Nx10

[IRLab @IRLab_UDC]

Hoy, en el #SaferInternetDay, reafirmamos nuestro compromiso: investigamos para detectar y frenar mensajes de odio y conductas abusivas, y para identificar señales de riesgo en redes sociales, con rigor y responsabilidad. Por un internet más seguro.

[Javier Parapar @jparapar]

🎉 Our paper “PartisanLens: A Multilingual Dataset of Hyperpartisan and Conspiratorial Immigration Narratives in European Media” accepted at #EACL2026 🇲🇦 main conference. Work done with Michele Maggini @itspalomap @anxopvila and the @HYBRIDS_MSCA team!
See you in Rabat! ✈️📍

[Javier Parapar @jparapar]

Happy to see this work with @davidoterof accepted at #ECIR2026. A deceptively simple yet effective idea: improve PRF through selective relevance feedback using small LLMs, denoising the feedback set without polluting or overcomplicating the underlying PRF model.