Fuzzy Matching Names and Addresses
Link records that are almost the same.
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
[VERIFY] is an editorial flag, not a blank to fill. Wherever the prompt shows [VERIFY], check that item against a reliable source before you use it.
Plan a fuzzy matching approach. Two lists with sample rows (anonymised): [SAMPLE_ROWS]. Volume of each: [VOLUMES]. The cost of a wrong match versus a missed match: [COSTS]. Outline a pipeline: standardise (case, punctuation, abbreviations), block to reduce comparisons (first letters, postcode), score candidate pairs with several string similarity measures, choose thresholds by labelling a sample, and review borderline cases manually. Provide Python code using standard libraries or common packages, with a note to check their current documentation. Show how to measure precision and recall on a labelled sample and how to set two thresholds for auto-accept and manual review. Warn about privacy: handle personal data according to the rules in [JURISDICTION] ([VERIFY]). Add a section on documenting the matching decisions so that they can be audited: store scores, the rule that accepted each pair, and the reviewer. Provide a small validation set design of 200 pairs, how to sample it fairly across score ranges, and a plan for revisiting thresholds when the data changes.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.