Dealing With Names and Addresses Around the World
summary
Names and addresses break common schemas: parsing first/middle/last fails for patronymics, single‑name cultures, different orderings and transliterations, and Western heuristics miss many cases. Practical specifics: use nominally for Western name heuristics, use pipal/libpostal to parse and normalize international addresses, rely on fuzzy matching for tolerant lookup, and avoid regular expressions for international validation. Commercial KYC or record‑linkage products cover harder script and cross‑language edge cases.
AI-written summary · about
chapters (17)
- 0:00Global naming challenges
- 1:26Speaker background
- 4:18Why names matter
- 5:22Parsing names into schemas
- 8:54Naming conventions worldwide
- 13:32Flaws in name parsing
- 14:13When name parsing is needed
- 17:27Python for structured names
- 21:44Open-source name tool limits
- 22:48Global address inconsistencies
- 26:48Libraries for address parsing
- 27:50Key takeaways
- 29:15Q&A: Badge sorting
- 31:22Q&A: Probabilistic name matching
- 33:09Q&A: Structuring unstructured data
- 34:35Q&A: Common names in searches
- 35:40Q&A: Matching signatures to identities
related talks
40mThe Eternal Pain of Accurate Addresses
05mPeople Suck a Naming Places
15mbeyond the names - A social trend analysis of name themes
45mCustomizable probabilistic record linkage with Name Match
33mAdd geocoordinates to your addresses in PostgreSQL
11mWhere? The gap between sensor data and humancentric geography in the AI age
25mRecursion Schemes Without the Barbed Wire (Part 5)
29mA Graph Entity Resolution Playbook