What's in a name? Phonetic Algorithms for Search and Similarity
summary
Phonetic hashing actually finds misspelled names: soundex and Postgres built-in functions, NYSIIS and Metaphone variants, and an elastic search phonetic plugin are compared with tradeoffs. A Lost-and-Found prototype for a 35,000-attendee convention that processes 1–2,000 items per weekend stores phonetic hashes and runs runtime translations to match typos; it works for many English variants but fails on most non‑English and many Unicode names. Caveats include Postgres soundex built-in limitations and operational pain running elastic search.
AI-written summary · about
chapters (13)
- 0:00Introduction to search
- 1:14How search engines work
- 2:49Database search techniques
- 5:08Full-text search tools
- 6:53Phonetic algorithm basics
- 8:45Understanding Soundex
- 12:20Soundex limitations
- 13:06The NYSIIS algorithm
- 15:19NYSIIS in practice
- 15:56Metaphone algorithms
- 16:58Implementing fuzzy search
- 21:28Non-English and future work
- 23:40Conclusion
related talks
33mFuzzy Search Algorithms How and When to Use Them
13mUnconference Session: A Neat Trick for Fuzzy Search
43mFull-Text-Search Explained from A to Z with French Food
09mFailed Searches
26mOn The Look-Out For Your Data
14mSearch on Steroids
59mFull-text search - seek and ye shall find
20mMultilingual Search System - How to Build and Improve!?