README.md

Description Generator Module

This module was designed with two strategies in mind:

• Use the input and manually return a response only by filling in a template with the values.

• Use a transformer to create a text given the input in such a way that it sounds human.

The second one will be left for future work since for the moment, the input is too small and a generative or summary model is not necessary.

Expected Input

For the input the model expect the folllowing dictionary/json:

{
    'author_name': 'Leonardo da Vinci',
    'art_name': 'Mona Lisa',
    'type': 'painting',
    'style': 'Renaissance',
    'objects': ['painting', 'Oil Paint'],
    'period': 'Renaissance',
    'date': '1503'
}

This is an example of what is expected, each field is described below:

• author_name The name of the creator
• art_name The name of the art piece
• type Type of work. (e.g. painintg, mixography, sculpture, etc)
• style The Style. (e.g Renaissance, Contemporany)
• objects The structure of the workpiece. The better for the descriptor output in case that the art_name and author is unknown.
• period The period
• date The date, can be exact or aproximate.

The previous fileds are string type and can be empty, but the less empty ones the richer will be the descriptor's output :)

Expected Output

{
    'description': 'some text'
}