Add more languages

[rashmiranjanrrs]

Hey can you add more Indic language or can you share the pattern or the structure of subset so that I can able to add new languages as per my requirement. How to add new subset ?

[landrok]

Hey,

How to add new subset ?

• clone repository
• create your subset file in src/LanguageDetector/subsets/ folder
• write at least one test in tests/LanguageDetector/LanguageDetectionTest.php file to validate your subset
• then you can push with a commit message Add new language {the new language}

Subset structure

A subset file is a JSON encoded file with the following structure:

{
  "freq":{"D":662077, [...], "tha":240340},
  "n_words":[260942223,308553243,224934017],
  "name":"en"
}

• freq contains a list of key => value pairs where key is the ngram and value is an integer that represents the number of occurences found in source files.
  LanguageDetector accepts unigrams, bigrams and trigrams.
• n_words is a serie of 3 integers that represents total number of occurences ordered by ngram size (1,2,3)
• name is the name of the language

More

A you may guess, a "learning" tool has to be written to generate a subset. It's not yet packaged with the library but might be in the future.
An advise: to generate a reliable subset file, you have to collect a large number of files in the desired language and, if possible, from various language variations.

Hope this helps

[devope]

@landrok hey! can you give any advice how to extend your library to support georgian (ka) language (https://en.wikipedia.org/wiki/Georgian_language)