Hi, i would like to use Morpho.Feature.POS but I am not sure how to implement it. Could someone please point me in the right direction?

• Conjugation
• Derivation
• Stemming

Changing the code and pushing it affects the test website behavior (Also, Github doesn't detect this project ass javascript but HTML: (I don't like that 👎 )

• When re-factored morse special chars; some punctuations in Arabic and Japanese get lost.
• Also, the Jpn char ん has the same morse cose as +
• The japanese space has some issues

Create the function:
A conjugation method which doesn't rely on any database. Even if the results are not 100% relevant. It can have a little table or something for irregular verbs.

Check JapKatsuyou project:
https://github.com/anqafalak/japkatsuyou

Charsets

やっぱり --> yaっpari

• README Add more to the project description
• Check the API documentation (add more stuff)
• change CREDITS.md to CONTRIBUTING.md (Github understand it that way) : https://github.com/kariminf/jslingua/community
• add a CODE_OF_CONDUCT whatever that is
• add coding specifications (more helpful than code of conduct)

The problem:

Not every language has the same conjugation pattern, for instance:

This is an example of tenses; but if we incorporate the mood, we will have: Conditional, Indicative, etc. and if we introduce the aspect, we'll have perfect, continues, simple, perfect continuous. Then the negation and the voice (active, passive). For other languages, like Japanese: Formality (plain, polite, formal).

Japanese doesn't need personal pronouns in its verb conjugation.

I think it is time to have a project icon 👍

• Porter stemmer [WIP: code quality].
• Lancaster stemmer.

It will be great if there is a default stemmer designed for JsLingua.

• Arabic
• English
• French
• Japanese

Not so much to do; mostly the same as English

Logistic regression (multiclass)

Will be used for MEMM (PoS tagging)

Will be used for parsing

When i create a new stemmer and recreate second one the second crushed the 1st stemmer

    Morpho.newStemmer.call(this, "porterStemmer", "English proter stemmr", porterStemmer);
    Morpho.newStemmer.call(this, "lancasterStemmer", "English Lnacaster stemmer", lancasterStemmer);

How to add new stemmer to English Morpho module without crushed previous stemmers ?
you'll find the code in #35

remain -> becomes "main" in the past. "I mained"
request -> "I quested"
return -> "I turned"

Also:
finish -> "I inish", "I inished"

Arabic Language detection:
isArabic(text)

Verb --> Noun
Verb --> adverb

etc.

Just check this:

https://noisy-text.github.io/2016/pdf/WNUT17.pdf

Languages like Arabic use some normalization in Information retrieval tasks. For example, the diacritics are removed, etc.

• We'll inspire them from nltk tokenzations

I prefer those based on regular expressions (rule-based):

I don't care if the char-by-char methods are more fast; because in the end, sending the task to user side will be more profitable than using a server-side (the exchange time)

Use a standard translateration:

• Hepburn romanization
• Kunrei-shiki romanization (ISO 3602)
• Nihon-shiki romanization (ISO 3602 Strict)

Conjugation of some irregular verbs is wrong: past participle

For example: override

Must slice the prefix over- before verifying the verb

See this https://en.wikipedia.org/wiki/Romanization_of_Arabic

• ALA-LC : used in scientific publications (I used it myself in an unpublished paper)
• chat : difficult since everyone uses its own chat marks; but useful in Information retrieval, extraction and understanding

Group 3 is not yet handled

Complete the task:
A conjugation method which doesn't rely on any database. Even if the results are not 100% relevant. It can have a little table or something for irregular verbs.

Force using MEMM for all languages

Then, each language must only implement words encoding ==> features extraction

Add French support for the available modules

Improve it. Look at this algorithm:

http://www.learnarabiconline.com/broken-plurals.shtml

Mostly it is done using patterns rather than rules. Never found a resource (either software or book) that uses rules.

There are some pattern verbs and a dictionary of verbs linked to these verbs expressing that they have the same conjugation pattern.

Information like: population, countries and dialects, is changing with time. So, this information can't be afforded by the module itself.

Mark the three functions as obsolete; delete their documentation and test.

Just add abbreviated functions for those who love short code

When we want to say 103000; it will be:
مائة وثلاثة آلاف
, but we get:
مائة وثلاثة ألف

• Singular to plural
• Masculine to Feminie

Arabic Romanization

• Buckwalter (from, to)
• etc.

Complete the task:
A conjugation method which doesn't rely on any database. Even if the results are not 100% relevant. It can have a little table or something for irregular verbs.

• Create modules for each type
• a function to detect the type of addition: ist, in-, etc.
• fix voice and negation for some types

Syntax level processing: define its API

Some languages like Japanese use often pairs of characters like "ka", "ki", etc.

The lists aforded by Shereen Khoja for week middle waw and yaa have some similar verbs.

For example: طار
it exists in waw and yaa

It should be just in yaa

So, the list must be revised

Just arrange it as:

Constructor:

Data:

Groups of functionalities

I've never found one (working alone, without a tokenizer)
I found this: https://github.com/miaoski/stemtokstop but the stemmer is actually the tokenizer itself

I will check KuruMoji project:

https://www.elastic.co/guide/en/elasticsearch/plugins/current/analysis-kuromoji-stemmer.html

http://atilika.org/kuromoji/

Most modules has complex method

CodeFactor found an issue: Complex Method

It's currently on:
jpn\jpn.morpho.js:322-495
Commit bc2b57f