Synopsis

Amber is a simple node.js framework for making personal assistants.

Motivation

Amber provides many abstractions to help developers think about their bots. I made Amber to provide a high level framework for working with personal assistants in a way more similar to working with graphs.

Usage

Here are some of the main abstractions in Amber:

Targets

A target is the action that the user wishes to perform with a query. Text is entered into a chatbot and a target parser (I have used, for example, Microsoft’s LUIS.) categorizes the text into one of many targets. A target is a thing that your bot can do.

Requirements

Sometimes to do something, your bot requires information about the user or the outside world. A requirement is something that your bot can have, like a location or a name. When defining a target, you can put in a list of requirements that must be fulfilled before the target is reached. Also, requirements can have requirements themselves. For instance, a label may be needed to find out the ID of a resource.

Filters

“What is the weather in LA” is clearly a weather target, but it also contains information about the location. Filters are defined with each requirement, and they take what we think might have enough information to fulfill a requirement and either fulfills is by transforming the information into the correct format, or tells the system that it still needs to ask the user about it.

Example

Here is a look at the graph for a bot I made which lets members check out resources for a VR Club: VR-Bot.

As you can see, to get to the target add_equipment, you have to first fulfill the requirements of type, owner, location, and label.

Let’s say we want to add a new Oculus Rift. We type “I want to add a new rift.” First, this sentence is passed to Microsoft’s LUIS. It returns:

{ “query”: “I want to add a new rift.”, “intents”: [ { “intent”: “add_equipment” } ], “entities”: [ { “entity”: “rift”, “type”: “type” } ] }

The target is set to add_equipment and Amber will try to fulfill the type requirement with “rift”. For type, the filter essentially finds the highest scoring type from the server using a modified version of this algorithm and fulfills the type requirement with an object containing the name of the type and its ID on the server.

Next, Amber will recursively attempt to fulfill every requirement for add_equipment. In this case, add_equipment is defined with the following requirements: [‘type’, ‘owner’, ‘location’, ‘label’]. Type is already fulfilled, so it will call toGet on owner to get the owner’s name and ID, and do the same for location and label. Once every requirement is fulfilled, add_equipment’s onExecute method is called, the user is prompted to confirm the accuracy of the information entered and they can add the new equipment.

Installation

npm install amberchat

Documentation

Documentation is provided at /docs and are available online here.

License

MIT License

Copyright (c) 2016 Keiran Paster

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.