The (Swiss cheese) data conundrum...

on YouTube ↗23m

" summary

Swiss Re Institute builds C.H.E.E.S.E. (Company Hierarchy End‑2‑End Structure Extraction) to join internal and purchased datasets, claiming unique company identifiers are the missing key and that sourcing can take months or years. The pipeline uses NLP standardization, TF‑IDF vectors, geocoding to lat/long, geocoding + bucketing for exact-match pruning and clustering to merge millions of records, outperforming simple fuzzy name matching and reducing duplicates to produce consolidated end‑to‑end risk views.

AI-written summary · about

" chapters (6)

1. 0:00Introduction
2. 3:08Data as insurance lifeblood
3. 6:41Data sourcing challenges
4. 12:04Insurers' data maturity
5. 14:38Company matching algorithm
6. 21:56Conclusion and key takeaways

" related talks

• 04mInterviewRashmi Shamprasad · Scale By The Bay 2019
• 09mRethink How You See DataSharmila Shahani Mulligan · Strata Conference New York + Hadoop World 2013
• 37mData Swamp to Data Swimming Pool: Cleaning up the MessJohn Joyce, Maggie Hays · Crunch Conference 2022
• 59mThe Great Data DebateDoug Cutting, Dan Wolfson, Phillip Radley, Raghu Ramakrishnan, Mike Ferguson · Big Data LDN 2018
• 29mIdentity Resolution – The Brain of Customer Data Platform (CDP)Upendra Singh · Great International Developer Summit 2025
• 50mData Privacy Concerns - A Risk Based Approach to De-IdentificationNoopur Davis, Doina Iepuras · RSA Conference 2023
• 07mData IntelligenceVaibhav Nivargi · Strata Conference New York + Hadoop World 2013
• 47mWhat it takes to process healthcare data?Piotr Kaczor · Confitura 2023