ingredient-parser-nlp 0.1.0a1

A Python package to parse structured information from recipe ingredient sentences

Ingredient Parser

The Ingredient Parser package is a Python package for parsing structured information out of recipe ingredient sentences.

1 large onion, finely chopped

becomes

{
    "quantity": 1,
    "unit": "large",
    "name": "onion",
    "comment": "finely chopped"
}

Documentation

Documentation on using the package and training the model can be found at https://ingredient-parser.readthedocs.io/en/latest/.

Quick Start

Install the package using pip

python -m pip install ingredient-parser-nlp

Import the ```parse_ingredient`` function and pass it an ingredient sentence.

>>> from ingredient_parser import parse_ingredient

>>> parse_ingredient("3 pounds pork shoulder, cut into 2-inch chunks")
{'sentence': '3 pounds pork shoulder, cut into 2-inch chunks',
 'quantity': '3',
 'unit': 'pound',
 'name': 'pork shoulder',
 'comment': ', cut into 2-inch chunks',
 'other': ''}

# Output confidence for each label
>>> parse_ingredient("3 pounds pork shoulder, cut into 2-inch chunks", confidence=True)
{'sentence': '3 pounds pork shoulder, cut into 2-inch chunks',
 'quantity': '3',
 'unit': 'pound',
 'name': 'pork shoulder',
 'comment': ', cut into 2-inch chunks',
 'other': '',
 'confidence': {'quantity': 0.9986,
  'unit': 0.9967,
  'name': 0.9535,
  'comment': 0.9967,
  'other': 0}}

The returned dictionary has the format

{
    "sentence": str,
    "quantity": str,
    "unit": str,
    "name": str,
    "comment": Union[List[str], str],
    "other": Union[List[str], str]
}

Model accuracy

The model provided in ingredient-parser/ directory has the following accuracy on a test data set of 25%:

Sentence-level results:
	Total: 9277
	Correct: 7689
	-> 82.88%

Word-level results:
	Total: 52931
	Correct: 50051
	-> 94.56%

Development

The development dependencies are in the requirements-dev.txt file.

Note that development includes training the model.

• Black is used for code formatting.

• isort is used for import sorting.

• flake8 is used for linting. Note the line length standard (E501) is ignored.

• pyrigt is used for type static analysis.

The documentation dependencies are in the requirement-doc.txt file.