nlp-compromise / compromise-plugin Goto Github PK

I want to do something like this to match mm/dd/yyyy, m/d/yyyy, mm-dd-yyyy & etc.

"patterns": {
"/^(\d{1,2})-|.|/-|.|/$/": "Date",
}

Are there any documentation explaining the syntax of this example?

//military-words
{
  plurals:{
    barracks:'barracks'
  },
  patterns:{
    "#Ordinal infantry? division":'Noun',
    "major (general|lieutenant|#Person)":'Person',
    "#Posessive six":'Noun',
    "over$":'Expression'
  },
  words:{
    'niner':'Value',
    'buck sergeant':'Person'
  }
}

Sorry this is a multi parts question

1. When I try to create a plugin with

  "patterns": {
    "to? (readme|instruction(s)?)": "Navigation"
  },

readme works as expected but can't get instruction or instructions to work.
However, this works

  "patterns": {
    "to? (readme|instructions|instruction)": "Navigation"
  },

2. Is plugin the right tool for creating new intents and actions?

3. Now that I can see in the terms where instructions is tagged with "Navigation" from this sentence "give me instructions". What method or how can I check programmatically that the sentence in question is indeed related to "Navigation"? And how can I train it to respond with "http://mydomain/readme.html"?

I have added the ability to provide responses as part of the plugin. I want to have a chat with you before I submit a PR.