This is the very basic demo application. It is created to introduce you the main concepts, show you files structure and explain you how to start working with Dasha.
Every application made on Dasha platform consists of two parts:
- Dialogue part
- contains:
- dialogue logic (
.dsl
files) - nlu dataset (
.json
file) with intents and entities - phrasemap (
.json
file) - main config (
.dashaapp
file) that combines all together and defines additional nlu skills and dsl entry point
- dialogue logic (
- running remotely on our servers
- contains:
- SDK part
- is written on Dasha SDK (
JavaScript
for now) - puposes:
- deploy the Dialogue part of application on server
- configue the whole application and single conversation instances
- provide dialogue inputs
- collect dialogue outputs
- integrations with backend
- implementation of external functions used in dialogue
- running localy on your machine
- is written on Dasha SDK (
Observation of the files presented in this demo.
app/app.dashaapp
- main app file - configures app with main dialogue file, phrasemap and nluapp/main.dsl
- file with dialogueapp/phrasemap.json
- phrasemap file - empty in this exampleindex.js
- sdk file, deploys and runs the app and executes the single conversation with cli argumentphone
(orchat
)package.json
- javascript config file
There is no dataset file in this demo.
The dialog specified in main.dsl
simply awaits any sounds from user (with 2000 ms timeout), then says "Hello" and awaits for any user input. After that Dasha says "Goodbye" and returns object {success: true}
.
- Setup environment (see doc)
- Run
npm i
in a current folder.
Run npm start chat
for launching text chat or run npm start <your_phone_number>
to start a phone call.
AI: Hello!
USER: abacaba
AI: Goodbye!