language-model-converter

This is a tool to convert language files defined in yaml language model into the LUIS internal JSON representation.

The language model

This is a yaml format file that is conveted to a JSON file that LUIS can undestand

Words preceed by # are comments, and its purpose is to leave some insights to people that will help on understanding whats going on. These words will be removed when training luis, so feel free to add as much as you want to help next colleagues that will manage the file

list.${examples}:  # Defines a example list to make substitutions in the utterances
  - heaven
  - hell

tef.intent.info: # Defines a Luis Intent
  - Tell me about the purgatory # Simple utterance
  - What is ${examples} # Substitution with list: Will generate for you "What is heaven" and "What is hell"

tef.intent.go: # Defines a Luis Intent
  - Go to [purgatory:tef.places] # Defines an entity "tef.places" giving "purgatory" as an example. The example is mandatory
  - Head to [${examples}:tef.places] # You can make substitution in the entity examples too!
  - We will go to [hell:tef.places] from [january 1:tef.date::start] until [december 31:tef.date::enf] # Hierarchical Entities are supported also

phraselist:
  aksforinfo: # ex: How to ask for something synonims:
    words:
      - Tell me about
      - What is

  tef.places: # ex: Place examples here for training entities
    words:
      - heaven
      - hell
      - purgatory
      - home
  # if you want no have a not-exangeable phraselist, add the property `mode: false` to it. Example:
  # tef.relations
  #   mode: false
  #   words:
  #     - leader
  #     - area
  # more info: https://github.com/Microsoft/Cognitive-Documentation/issues/97#issuecomment-265738124

builtin: # LUIS builtin entities that should be used
  - age
  - datetime
  - dimension
  - encyclopedia
  - geography
  - money
  - number
  - ordinal
  - percentage
  - temperature

The list of builtin entities can be found in the LUIS Help and/or the Cognitive Services Help Please, read carefully about what entities are available for the target language you are writting utterances to, as

Unless otherwise noted, each pre-built entity is available in all LUIS application locales (cultures).

Note: maybe the doc page is not actualized with all the entities

You dont need to declare with brackets the builtin entities. They are recognized by default on LUIS. I.E:

# DONT DO THIS
tef.intent.money:
  - It's worth [$30:builtin.money]

# DO THIS
builtin:
  - money

tef.intent.money:
  - It's worth $30

Limitations

As of writing this doc, the following limitations apply in LUIS Service

• Max of 40 intents (tef.intent.info + tef.intent.go = 2)
• Max of 50 chars for the intent name (tef.intent.info = 15)
• Max of 10 phraselists (askforinfo + tef.places = 2)
• Max of 10 entities ([something:tef.places]) You can provide as much examples as you want for the entities
• Max of 10 children entities per parent ([something:tef.date::start])

Usage examples

npm install -g @telefonica/language-model-converter
language-model-converter --help

  Usage: language-model-converter [options] <files>

  Convert language files defined to LUIS format

  Options:

    -h, --help               output usage information
    -c, --culture <culture>  Culture code this files belongs to (ex. "en-us")
    -n, --ner                Generate a model compatible with NER

  Examples:

    Convert all files in 'models' and its subfolders, starting with 'en',
    setting the locale to en-us
      $ language-model-converter ./models/**/en*.yaml -c en-us

Notice that you can use luis-cli to import or update the generated LUIS models.

USE IT PROGRAMMATICALLY

You can also use this module from code:

import { LanguageModelParser, culture } from '@telefonica/language-model-converter';

let parser = new LanguageModelParser();
let files = 'model.yaml';
try {
    let luisModel = parser.parse(files, 'en-us' as culture);
    // Do something with the LUIS model
} catch (err) {
    // Manage errors
}

LICENSE

Copyright 2016 Telefónica I+D

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.