professor - university of amsterdam. thinking: data, links, remixing, knowledge, provenance, espresso. My opinions. Mastodon: @pgroth@sigmoid.social

Amsterdam
Joined July 2009
Raymond Chandler 1944 on research communication – at Amsterdam, The Netherlands
2
8
2
26
Paul Groth retweeted
📣 Join us at TRIPLET 2026 @eswc_conf in Dubrovnik! 🌟 Keynote by @pgroth ! 🏆 TRIPLET Challenge presentations 🥇 Leaderboard winners announced for Task 1 and Task 2 📅 11 May 2026 🔗 ecladatta.github.io/triplet2… #TRIPLET2026 #ESWC2026 #SemanticWeb #KnowledgeGraphs #NLP #MultimodalAI
2
3
325
Also the Satellite proceedings for #ESWC2024 are out! Thanks to the whole team @eswc_conf and the General Chair @albertmeronyo Part I: link.springer.com/book/10.10… Part II: link.springer.com/book/10.10… #semanticweb #computerscience #knowledgegraphs #llm
1
6
11
617
Paul Groth retweeted
📢 New course alert 📢 I am currently teaching a course on "Language Models and Structured Data" at Institut Polytechnique de Paris. Topics: Language Models, LoRa, Quantization, RAG, Graphs, Tabular Data, Text2SQL Zenodo: zenodo.org/records/14673347
2
11
569
Buckle up because we're crashing into the new year with my annual database retrospective: License change blowbacks! @databricks vs. @SnowflakeDB gangwar! @DuckDB shotgun weddings! Buying a college quarterback with database money for your new lover! cs.cmu.edu/~pavlo/blog/2025/…
18
161
28
723
88,084
In our new @PNASNews paper, across 21 experiments with 23,000+ participants, we identify a critical distortion that shapes decisions involving tradeoffs: we find that people systematically overweight quantified information in such decisions. Paper: pnas.org/doi/10.1073/pnas.24… 🧵
4
27
3
105
20,649
We have an opening for a PhD student investigating concept drift in sensor rich environments. Come work with the awesome @vdegeler werkenbij.uva.nl/en/vacancie…
3
3
420
Really proud of @James_G_Nevin - a fantastic PhD student. Was fun to supervise him together with @mhlees . We know that data handling (i.e. data integration, cleaning, etc) can have lots of downstream impacts. Here's evidence.
Congratulations to Dr. @James_G_Nevin who successfully defended his PhD thesis The Ramifications of Data Handling for Computational Models. Check it out: hdl.handle.net/11245.1/d3da6… A collaboration with @UvA_CSL in the @UvA_IvI co-supervised @mhlees @pgroth
3
278
Paul Groth retweeted
Brilliant and engaging talk by Teresa Liberatore at #EKAW2024: Influence Beyond Similarity—A Contrastive Learning Approach to Object Influence Retrieval. Insightful ideas and impactful research!
6
12
567
We're at #EKAW2024 this week across the street @CWInl . We have two papers: one on object influence retrieval & the other on the impact of entity linking. We also have multiple workshop contributions as well. Info at: indelab.org/news/ @ekawconference
1
4
274
Tomorrow the #EKAW2024 party will really kick off, but the proceedings are already available! Click on the link at event.cwi.nl/ekaw2024/ for temporary free access! #semweb #knowledgeengineering #knowledgemanagement #languagemodels #conference @ekawconference
1
4
1
19
1,012
Fascinating talk by @ioanamanol on dealing with all the data models for data journalism at @iswc_conf #ISWC2024 Very cool use of gittables to retrieve names for entities (cc @MadelonHulsebos)
15
657
We're excited to be at #iswc2024 this week. Come talk to us about our work on knowledge graphs, LLMs, tables, and knowledge engineering: @pgroth @bradleypallen @LiseStork @JanCKalo . @iswc_conf @lm_kbc
2
8
549
This is the best paper written so far about the impact of AI on scientific discovery
Readers added context they thought people might want to know
There seem to be grave concerns about the integrity of the research. MIT announced that they "conducted an internal, confidential review and concluded that the paper should be withdrawn from public discourse." economics.mit.edu/news/assuring-… wsj.com/tech/ai/mit-sa…
103
1,551
299
7,600
5,680,086
🚨 What’s the best way to select data for fine-tuning LLMs effectively? 📢Introducing ZIP-FIT—a compression-based data selection framework that outperforms leading baselines, achieving up to 85% faster convergence in cross-entropy loss, and selects data up to 65% faster. 🧵1/8
10
44
7
242
46,212