zer-blocking
Blocking strategies and inverted index for candidate pair generation in the zer entity-resolution library.
Blocking drastically reduces the number of record pairs that need to be compared by grouping records into blocks; only pairs within the same block are compared. This crate provides a composable, pluggable set of blocking keys, with built-in keys optimized for Dutch administrative data.
- Documentation: docs.zal-analytics.ch
- Website: www.zal-analytics.ch
- Support & feedback: info@zal-analytics.ch
What it provides
| Item | Description |
|---|---|
CompositeBlocker |
Combines multiple blocking keys; a pair is a candidate if it matches on any key |
InvertedIndex |
Efficient inverted index mapping blocking keys to record IDs |
BlockerFactory |
Builds a CompositeBlocker from a Schema automatically |
SchemaCategory |
Enum describing the dataset type (BRP, KvK, ANPR, …) |
Built-in blocking keys
| Key | Description |
|---|---|
| Phonetic | Soundex / Double Metaphone on name fields |
| Date | Exact-match on date fields (birth date, registration date) |
| Suffix | Last-N-characters of string fields |
| Alias | Normalised alias / nickname matching |
| Transliterated | ASCII transliteration of accented Dutch characters |
| Vehicle | Licence plate normalisation for ANPR data |
License
Apache-2.0 · GitHub