What is the Adapt Intent Parser?

The Adapt Intent Parser is open source software. It is lightweight and is designed to run on devices with limited computing resources, such as embedded devices.

Adapt takes in natural language as an input, and outputs a data structure that includes

• the Intent: what the User is trying to do

• a match probability: how confident Adapt is that the Intent has been correctly identified

• a tagged list of entities: that can be used by Skills to perform functions

Adapt is useful for interpreting natural language input. For example, you might want to create a voice user interface that allows a User to play a Pandora station. The Utterances a User might Speak include:

• Turn on Pandora

• Play Pandora

• Play my Joan Jett Pandora station

The Adapt Intent Parser takes this input and generates a JSON data structure like this:

{
    "confidence": 0.61,
    "target": null,
    "Artist": "joan jett",
    "intent_type": "MusicIntent",
    "MusicVerb": "put on",
    "MusicKeyword": "pandora"
}

Applications - or Mycroft Skills - can then parse the JSON data take appropriate action - such as playing Joan Jett using the open source Pandora application Pianobar.

Is Adapt right for me or my use cases?

Adapt is useful for determining user intent quickly and accurately from natural language. It is particularly beneficial for applications with limited vocabularies - that is, where the User will only issue a small number of Utterances.

Adapt has a number of features.

Lightweight

Adapt is written to run on embedded system with limited resources. Adapt can be used as the engine to parse natural language on IoT devices like remote controls and hubs.

Portable

Adapt is written in Python. Python is very widely adopted and supported in the technical community. Because it is written in Python, Adapt will run on many different devices - from servers to a Raspberry Pi. Adapt is designed to be run cross-platform and can be deployed in software environments that include Android, iOS, Windows and Linux.

Reliable

Because Adapt is small enough to run locally on an embedded device, it removes the dependency of having to parse intent in the cloud. This allows applications to function when they don't have an internet connection. Adapt is therefore suited to environments where the internet connection is slow, unreliable or intermittent.

Open source

At Mycroft.AI, we believe that artficial intelligence is too important to be controlled by a few large companies. Adapt is open source, licensed under the Apache v2.0 license. Anyone can access the source code, change the software and use Adapt Intent Parser in new applications.

​Open Source software powers the Internet. Artificial intelligence software, like Adapt Intent Parser, is gaining in importance. However, very few artificial intelligence software is open source. Here at Mycroft.AI, we believe that natural language processing is going to be a key component of many future technologies. By sharing our software we believe that we can help bring artificial intelligence to the open source community where it can be used to benefit millions of users world wide. We hope to work with other open source initiatives like TensorFlow and OpenAI to ensure that the future of artificial intelligence is open for all.

Who developed Adapt?

Adapt Intent Parser was developed at Mycroft.AI by a team led by Sean Fitzgerald. Sean was previously a developer at both Siri and Amazon Echo.

How do I install Adapt Intent Parser?

Requirements

Adapt Intent Parser requires Python 2.7.

pip installation

pip is a package manager used to install and manage software packages in Python. To use pip you first need to install it.

Installing pip for Ubuntu / Debian based Linux

$ sudo apt-get install python-pip python-dev

Installing pip for Fedora / RPM based Linux

$ sudo dnf install python-pip python-devel

Installing Adapt Intent Parser

NOTE: For development on Adapt and Mycroft we use the sandboxing tool virtualenv to isolate Adapt from other software. This allows for dependencies to be configured just for Adapt, rather than installing them globally. This approach prevents conflicts between Adapt and other software.

Installing Adapt for Ubuntu, Debian or Raspbian

Install Adapt via pip into your project's virtualenv:

$ sudo apt-get install virtualenv
$ virtualenv myvirtualenv
$ . myvirtualenv/bin/activate
$ pip install -e git+https://github.com/mycroftai/adapt#egg=adapt-parser

How do I contribute to Adapt Intent Parser?

For Ubuntu, Debian and Raspbian based Linux

To develop on Adapt itself, or try the examples, check out the repo from GitHub and set up a virtualenv.

$ sudo apt-get install virtualenv
$ virtualenv myvirtualenv
$ . myvirtualenv/bin/activate
$ git clone https://github.com/mycroftai/adapt
$ cd adapt
$ pip install -r requirements.txt

Verify the installation by running the example code from the Adapt repository.

$ cd ~/adapt
PYTHONPATH=. python examples/multi_intent_parser.py "play some music by the clash"

For Fedora / RPM based Linux

To develop with the latest Adapt, install it via pip into your project's virtualenv:

$ pip install virtualenv
$ virtualenv myvirtualenv
$ source myvirtualenv/bin/activate
$ pip install -e git+https://github.com/mycroftai/adapt#egg=adapt-parser

To develop on Adapt itself, or try the examples, check out the repo from GitHub and set up a virtualenv.

$ pip install virtualenv
$ virtualenv myvirtualenv
$ source myvirtualenv/bin/activate
$ git clone https://github.com/mycroftai/adapt
$ cd adapt
$ pip install -r requirements.txt

Verify the installation by running the example code from the Adapt repository.

$ cd ~/adapt
PYTHONPATH=. python examples/multi_intent_parser.py "play some music by the clash"

For MacOS X

First, Xcode must be installed and Apple’s license agreement accepted.

$ xcode-select --install

Next, install the Homebrew package manager and update brew packages.

$ ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

brew update

Next, install Adapt dependencies with brew and python-pip.

$ brew install git python
$ pip install python virtualenv

To develop with the latest Adapt, install it via pip into your project's virtualenv:

$ virtualenv myvirtualenv
$ . myvirtualenv/bin/activate
$ pip install -e git+https://github.com/mycroftai/adapt#egg=adapt-parser

To develop on Adapt itself, or try the examples, check out the repo from GitHub and set up a virtualenv.

$ virtualenv myvirtualenv
$ . myvirtualenv/bin/activate
$ git clone https://github.com/mycroftai/adapt
$ cd adapt
$ pip install -r requirements.txt

Verify the installation by running the example code from the Adapt repository.

$ cd ~/adapt
PYTHONPATH=. python examples/multi_intent_parser.py "play some music by the clash"