Wrong codes being generated

Question

Wrong codes being generated

andreialecu opened this issue 8 years ago · comments

I've been troubleshooting some issues with text matching in something I'm working on and realized that this library generates wrong codes.

For example alexander should encode to ALKS but this library encodes it as ALNTR. Also aleksander encodes to ALKSNTR. Matching is pretty much impaired by this.

Is this expected with this implementation?

Titus · Answer 1 · Thu Oct 27 2016 15:21:42 GMT+0800 (China Standard Time)

For example alexander should encode to ALKS but this library encodes it as ALNTR. Also aleksander encodes to ALKSNTR. Matching is pretty much impaired by this.

Do you have a source for this?

Titus · Answer 2 · Mon Oct 31 2016 05:33:25 GMT+0800 (China Standard Time)

@andreialecu Ping!

Andrei Alecu · Answer 3 · Thu Nov 03 2016 19:08:57 GMT+0800 (China Standard Time)

@wooorm Sorry, didn't see this.

The source is pretty much every other double-metaphone implementation that I have come across.

Also, I think that strings like alexander and aleksander should generate the same codes, otherwise phonetic matching would be pointless.

Here's what natural generates: https://runkit.com/58136f736c8dfa00144f2065/581b19b0bf25ea001464b5fc

The two match.

And here's what this package outputs:
https://runkit.com/58136f736c8dfa00144f2065/581b1a8b5f41420013e08436