Awesome

pyopenie

Python wrapper for OpenIE5. This simply wraps the API from the server included with OpenIE 5.1.0.

Install

pip install pyopenie

Usage

First make sure you have the OpenIE5 server running. See the instructions here for how to do that.

Then the setup just requires you to pass in the url of the server:

>>> from pyopenie import OpenIE5
>>> extractor = OpenIE5('http://localhost:9000')

Any English sentence can be passed to OpenIE5 server.

>>> extractions = extractor.extract("The U.S. president Barack Obama gave his speech to thousands of people.")

The result is a JSON list of extractions with confidence, offset and other properties.

>>> extractions
[
    {
        "confidence": 0.38089450366724514,
        "sentence": "The U.S. president Barack Obama gave his speech on Tuesday to thousands of people.",
        "extraction": {
            "arg1": {
                "text": "Barack Obama",
                "offsets": [...]
            },
            "rel": {
                "text": "[is] president [of]",
                "offsets": [...]
            },
            "arg2s": [
                {
                    "text": "United States",
                    "offsets": [...]
                }
            ],
            "context": null,
            "negated": false,
            "passive": false
        }
    },
    {
        "confidence": 0.9168198459177435,
        "sentence": "The U.S. president Barack Obama gave his speech on Tuesday to thousands of people.",
        "extraction": {
            "arg1": {
                "text": "The U.S. president Barack Obama",
                "offsets": [...]
            },
            "rel": {
                "text": "gave",
                "offsets": [...]
            },
            "arg2s": [
                {
                    "text": "his speech",
                    "offsets": [...]
                },
                {
                    "text": "on Tuesday",
                    "offsets": [...]
                },
                {
                    "text": "to thousands of people",
                    "offsets": [...]
                }
            ],
            "context": null,
            "negated": false,
            "passive": false
        }
    }
]

Individual properties can also be accessed.

>>> extractions[0]['confidence']
0.38089450366724514

>>> extractions[0]['extraction']['arg1']['text']
'Barack Obama'