/semantic

A Python library for extracting semantic information from text, such as dates and numbers.

Primary LanguagePythonMIT LicenseMIT

Semantic

Semantic is a Python library for extracting semantic information from text, such as dates and numbers. Full documentation is available on PyPI, with a list of primary features and uses-cases below.

Build Status

PyPI version

Installation

Installing semantic is simple:

$ pip install semantic

Features

Semantic consists of four main modules, each of which corresponds to a different semantic extractor. The test suite (test.py) contains tons of examples for each of the four modules, but some sample use-cases are described below.

Dates (date.py)

Useful for:

  • Extracting relative (e.g., "a week from today") and absolute (e.g., "December 11, 2013") dates from text snippets.
  • Converting date objects to human-ready phrasing.

Numbers (number.py)

Useful for:

  • Extracting numbers (integers or floats) from text snippets.
  • Converting numbers to human-readable strings.

Example usage:

#!/usr/bin/env python
from semantic.numbers import NumberService

service = NumberService()

print service.parse("Two hundred and six")
# 206

print service.parse("Five point one five")
# 5.15

print service.parse("Eleven and two thirds")
# 11.666666666666666

print service.parseMagnitude("7e-05")
# "seven to the negative five"

Math (solver.py)

Useful for performing mathematical operations expressed as words.

Example usage:

#!/usr/bin/env python
from semantic.solver import MathService

service = MathService()

print service.parseEquation("Log one hundred and ten")
# 4.70048

Units (units.py)

Useful for converting between units expressed as words.

Example usage:

#!/usr/bin/env python
from semantic.solver import ConversionService

service = ConversionService()

print service.convert("Seven and a half kilograms to pounds")
# (16.534, 'lbs')

print service.convert("Seven and a half pounds per square foot to kilograms per meter squared")
# (36.618, 'kg/m**2')

Dependencies

The Dates, Numbers, and Math modules can run in isolation (i.e., without any dependencies), while the Units module requires quantities and Numpy.

License

MIT © Charles Marsh