Analyzing the Bot

Listening to action event

to track chat events simply use

const { Router } = require('wingbot');
const botRoot = new Router();

botRoot.on('action', (senderId, action, text, req, lastAction, doNotTrack, skill, res) => {

    if (action) {
        // send info that `action` was passed
    }

    if (req.isText()) {
        // was pure test
    }

    // or use whole request event
    myMagicTracker(req.event);
});

Logging AI success rate

aiHandled property is set to true, when intent was matched (using ai.match() method or ai.navigate()). Otherwise it is set to false.

You can set req.aiHandled to true with method req.markAsHandled()

const { Router, ai } = require('wingbot');
const botRoot = new Router();

botRoot.use(ai.match('intent'), ...);

botRoot.on('action', (senderId, action, text, req, lastAction, doNotTrack, skill, res) => {

    // logging winning intent
    if (req.intents && req.intents.length > 0) {
        const [{ intent, score, entities = [] }] = req.intents;


    }
});

When you need to collect AI feedback from navigator/makeSure matchers use ai.onConfirmMiddleware(<handler>) middleware.

Tracking disambiguation

It's good to track when disambbiguation occurs and if user selects the right meaning

const {
    Router,
    Processor,
    FLAG_DISAMBIGUATION_OFFERED,
    FLAG_DISAMBIGUATION_SELECTED
} = require('wingbot');
const botRoot = new Router();

botRoot.on('action', (senderId, action, text, req, lastAction, doNotTrack, skill, res) => {

    if (senderMeta.flag === FLAG_DISAMBIGUATION_OFFERED) {
        const { disambiguationIntents } = senderMeta;
        // when bot offers disambiguationIntents to let the user to choose
    }
});

const processor = new Processor(botRoot);

botRoot.on('action', (senderId, action, text, req, lastAction, doNotTrack, skill, senderMeta) => {

    if (senderMeta.flag === FLAG_DISAMBIGUATION_SELECTED) {
        const { likelyIntent, disambText } = senderMeta;
        // when user means disambText is likelyIntent
    }
});