ngram

Go implementation of Kondrak N-SIM and N-DIST (SPIRE 2005). Built for Watchman token scoring.

import "github.com/PhonoGrams/ngram"

ngram.Similarity("toradol", "tegretol") // N-SIM, n=2 (BI-SIM)
ngram.NSim("cycloserine", "cyclosporine", 3)
ngram.NDist("precede", "preceed", 2)

First-letter affixing (the first character is repeated n-1 times). Positional n-gram credit is the fraction of matching character positions.

References

• Kondrak, N-Gram Similarity and Distance, SPIRE 2005
• Watchman: moov-io/watchman#882

License

Apache License 2.0