v3.1 with Coverage algorithm and increased performance
Changes
Requires .Net 8.0 for self-host
⚡️ Overall 20% faster indexing speed
IMDB Top 10.000 indexes in 53 milliseconds.
✅ General API changes for ease of use
🚀 Increased search speed for larger datasets
🪄 New Coverage algorithm
Detects tokens (words) and returns an index as an int to truncate result list. Truncates on a single or several words.
Detects exact string similarity, great for shorter strings
Complements the RelevancyRanking algorithm which handles spelling errors. Coverage truncates on exact spelling of tokens only, but supports for example one correctly spelled and one incorrectly spelled token. In that scenario it will truncate the list on the correctly spelled token, but lift the result with incorrectly spelled string on top.
Requires a TokenizerSetup to work
Known bugs and limitations
StringReplacer does not affect Coverage algorithm in this release.