SimGus / Chatette

A powerful dataset generator for Rasa NLU, inspired by Chatito
MIT License
318 stars 56 forks source link

Support for lookup tables #14

Closed dannyheard7 closed 5 years ago

dannyheard7 commented 5 years ago

https://blog.rasa.com/improving-entity-extraction/

If this library could take in a lookup file flag that maps to the lookup_tables json entry, it would be really useful

SimGus commented 5 years ago

I will look into that. This being said this is not a top priority (and I don't have much time at the moment), so I can't promise you to add support for that soon.

Anyway, thanks for your request :)

SimGus commented 5 years ago

Since v1.5.0, you can use the --base-file command line option to provide a base file to extend.

If you give it a file containing for example:

{
    "rasa_nlu_data": {
        "lookup_tables": [
            {
                "name": "company",
                "elements": "data/company/startups.csv"
            }
        ],
        "common_examples": null,
        "entity_synonyms": null,
        "regex_features": []
    }
}

the generated examples will be added in the common_examples (and entity_synonyms) fields of such an object. This is described here.

I don't think I will add a more complete support for lookup tables than that (at least not at the moment).

Sorry for taking so long to look into this, I've been very busy.