We found that the package provide a weak performance where many words may have the same hash. To improve the performance we replaced the fixed prefix 'x' into a maped version of the first letter. Tests show a remarkable improvement in the performance and the results as well. Moreover, we treated both ة and ه at the end of the word as a ه in both cases.
There are many libraries that generate soundex index for English but non for Arabic.
In this contribution we provide the first open source soundex index for Arabic.
Our Arabic soundex library could help in improving the search in arabic documents, finding and correcting spelling errors and many other applications.
We developed an algorithm based on Tajweed rules that categorize the characters into groups based on it's pronunciation.
The second algorithm inspired from this source
npm i ar-soundexMake sure that you have installed devDependencies or run:
npm iIf you have trouble installing rollup you may check it's documentation rollup installed
To build the package on your machine run the following command
npm run buildvar arSoundex = require('ar-soundex');
arSoundex('عبدالله') // => x74600| parameter | Type | Description | Defult |
|---|---|---|---|
| word | String | A string of one word or more | '' |
| length | Number | The max length of generated index | 6 |
| method | Number | To set which algorithm to be used [1 or 2] | 1 |
// Detecting misspelling
arSoundex('عبدالله') // => x74600
arSoundex('عبدلله') // => x74600
// Detecting extra charactersarSoundex('ضربوه', 6, 1) // => x26700
arSoundex('ضرب', 6, 1) // => x26700
All contributions are welcome. Create a pull request here
If you have an issue please let us know . Create an issue here
Arabic Natural Language Soundex Metaphone Index Search Spelling