MindsDB home pagelight logo
  1. NLP
  2. Extract JSON from Text Data

In this example, we use the OpenAI model to extract data in a predefined JSON format from the input text data.

Default Model

When you create an OpenAI model in MindsDB, it uses the gpt-3.5-turbo model by default. But you can use the gpt-4 model as well by passing it to the model-name parameter.

​
Example in SQL

Let’s create an OpenAI model.

CREATE MODEL mindsdb.nlp_model
PREDICT json
USING
    engine = 'openai',
    json_struct = {
        'rental_price': 'rental price',
        'location': 'location',
        'nob': 'number of bathrooms'
    },
    input_text = 'sentence';

We pass three parameters in the USING clause:

  1. The engine parameter ensures we use the OpenAI engine.
  2. The json_struct parameter stores a predefined JSON structure used for the output.
  3. The input_text parameter contains the name of the column that stores input text.

Now we can query the model, passing the input text stored in the sentence column.

SELECT json
FROM mindsdb.nlp_model
WHERE sentence = 'Amazing 3 bedroom apartment located at the heart of Manhattan, has one full bathrooms and one toilet room for just 3000 a month.';

On execution, we get:

+----------------------------------------------------------+
| json                                                     |
+----------------------------------------------------------+
| {"location":"Manhattan","nob":"1","rental_price":"3000"} |
+----------------------------------------------------------+

​
Example in MQL

Please check out our docs on how to connect Mongo Compass and Mongo Shell to MindsDB.

To create this model in MQL, run the below command from Mongo Compass or Mongo Shell:

db.models.insertOne({
    name: 'nlp_model',
    predict: 'json',
    training_options: {
        engine: 'openai',
        input_text: 'sentence',
        json_struct: {
            'rental_price': 'rental price',
            'location': 'location',
            'nob': 'number of bathrooms'
        }
    }
})

We pass the same three parameters here.

  1. The engine parameter ensures we use the OpenAI engine.
  2. The json_struct parameter stores a predefined JSON structure used for the output.
  3. The input_text parameter contains the name of the column that stores input text.

Now we can query the model, passing the input text stored in the sentence column.

db.nlp_model.find({
    'sentence': 'Amazing 3 bedroom apartment located at the heart of Manhattan, has one full bathrooms and one toilet room for just 3000 a month.'
    })

On execution, we get:

{
  json: {
    rental_price: '3000',
    location: 'Manhattan',
    nob: '1'
  },
  sentence: 'Amazing 3 bedroom apartment located at the heart of Manhattan, has one full bathrooms and one toilet room for just 3000 a month.'
}