InAI Advances BriefsbyGiuseppe Futia·1d agoThe Semantic Web Is in the Business of Trust — Notes from ISWC 2026A week in Bari (Italy) between keynotes, papers, and my own talk on biomedical ontology alignment
Steven Renwick·5d agoElon Musk gave $347 million in 15 weeks. I tried to add it up.The headline numbers were easy to confirm. Working out what counts as “his” money took four different names and a lot of matching.
Steven Renwick·5d agoTwo very different money machines are fighting over one Texas Senate seatJames Talarico has more than 120,000 named donors. Ken Paxton has about 17,000, and a small circle of people who write very large cheques…
Tilo Renachowski·5d agoReal-Time Entity Resolution Is Not a Faster Batch JobA Kafka and Spark pipeline can ingest every event as it happens and still hand a live query a stale answer. Here is where the two clocks…
Ankit Agarwal·Oct 29Person Identity Resolution: The Model Matters Less Than How Many Pairs You Ask It to JudgeWe tested a self-hosted decision model SemIf against Databricks’ ai_decide() for healthcare members’ identity resolution use-case. Here is…
InTiloresbyTilo Renachowski·Oct 26A One-Off Clean-Up Starts Decaying the Day It FinishesA bank's migration clean-up is right on cutover day and a little wrong every day after. What to test, and what to put in the budget, so…
InTiloresbyTilo Renachowski·Oct 26‘The Threshold Was Above 90%’ Is Not an ExplanationWhat a supervisor, an auditor or a customer is owed when two records are merged, and why a score cannot supply it.
Ayush Raj·Oct 8How We Finished Top 50 of 32,000+ Teams at the Amazon ML Challenge 2026Entity resolution, open-set recognition, and a country we’d never seen, explained from scratch
Steven Renwick·Sep 26The Wrong Customer Merge Is a Privacy IncidentA wrong customer merge is not a harmless data-quality defect. The fix is reconstruction-based: preserve the records, evidence, policy…
