convo-platform

Small nodejs package that will (hopefully) make building chatbots easier.

Usage no npm install needed!

<script type="module">
  import convoPlatform from 'https://cdn.skypack.dev/convo-platform';
</script>

README

convo-platform

Small nodejs package that should (hopefully) make building chatbots easier.

Installation

This is a Node.js module available through the npm registry.

Before installing, download and install Node.js. Node.js v8 or higher is required.

Install the package using the npm install command:

$ npm install convo-platform

Features

  • State definition
  • State import/export

Quick Start

Firstly, you have to create a new Chatbot instance. Provide a function that processes a sentence (some form of NLP) and returns an object with {utterance, intent, score} properties.

const { Chatbot, ChatbotState } = require('convo-platform');

const bot = new Chatbot(yourNlpFunction);

Then, you have to define at least one new ChatbotState. Give it a name and add some answers for intents.

const firstState = new ChatbotState("state1");

firstState.addAnswer("greeting.hello", "Greetings friend.");
firstState.addAnswer("greeting.hello", "Hello user, how are you?");

firstState.addAnswer("feeling.positive", "That is nice to hear :)");

firstState.addAnswer("feeling.negative", "Sorry to hear that :(");

firstState.addAnswer("greeting.bye", "Goodbye!");
firstState.addAnswer("greeting.bye", "See you later.");

You can also add one action per intent (adding multiple actions for same intent will result in previous actions being overwritten). If you want to add multiple actions, add an anonymous function that calls all of them. The action will take the current ChatbotState as its first and only parameter.

firstState.addAction("greeting.hello", initializeSomethingFunction);
firstState.addAction("feeling.positive", (state) => {
    callFirstFunction("do this");
    callSecondFunction("do that");
    callThirdFunction(state);
});

After defining your state, don't forget to add it to the Chatbot (you can do this immediately after creating the ChatbotState as well, since only the reference is passed).

bot.addState(firstState);

Then, when the bot is set up, you can simply call the process function with your sentence to retrieve the result. This function is asynchronous, so it returns a Promise. The Promise resolves into an object with {utterance, intent, score, answers} properties.

// using Promise syntax
bot.process("Hello bot")
    .then((result) => {
        console.log(`Recognized intent "${result.intent}" from "${result.utterance}" with ${result.score} certainty.`);
        for (let i = 0; i < result.answers.length; i++) {
            console.log(`Bot answer [${i}]: ${result.answers[i]}`);
        }
    })
    .catch(console.error);

// inside async function
(async () => {
    let result = await bot.process("Hello bot");
    console.log(`Recognized intent "${result.intent}" from "${result.utterance}" with ${result.score} certainty.`);
    for (let i = 0; i < result.answers.length; i++) {
        console.log(`Bot answer [${i}]: ${result.answers[i]}`);
    }
})();

This should result in all answers for the recognized intent being logged in the console. Note that the result.answers contains all defined answers for the recognized intent. If only one answer is defined, it is still an array, albeit with only one element. If no answers are defined for the recognized intent, the result.answers array is empty.