CHATBOT.md

Host your Chatbot on Docker

This is the guide for preparing Bluemix and Docker environments

Download docker branch of this repository

• Click here to download the docker branch

• After downloading, unzip file Chatbot-docker.zip

  

Acquire Visual Recognition service on Bluemix

• After register the Bluemix account, then sign in

• Go to Bluemix catalog, search Visual Recognition

  

• Click Create button after selecting the Free plan

  

Get API key from tab of Service credentials then apply it in /Chatbot-docker/docker/Dockerfile.

• Click on Service credentials tab, then select credential dropdown button, copy api_key

  

• Update /Chatbot-docker/docker/Dockerfile, paste copied api_key to Dockerfile

  

• Please be aware of the variables of the values DO NOT HAVE quotes

  • Good example:

     ENV VISUAL_RECOGNITION_API_KEY abcdefghijklmnopqrstuvwxyz12345678901234

  • Bad example

     ENV VISUAL_RECOGNITION_API_KEY "abcdefghijklmnopqrstuvwxyz12345678901234"

Train Visual Recognition API with the training files, get new trained classifier ID then apply it in /Chatbot-docker/docker/Dockerfile

• Go back to Bluemix console, go back to Manage tab then click on Visual Recognition Tool (Beta) button to launch the Visual Recognition Tool

  

• Due to the GFW blocked the unsecured connection of Bluemix as below, use https://visual-recognition-tooling.mybluemix.net/

  

• For the first time visit of this tool, the API key is needed for authentication

  

• Then click on Create button to get started

  

• You will see this page, go to next step refer to following step

  

Download training images and unzip it

• Download training files here

• Unzip the Chatbot-training.zip file to /Chatbot-training

  

Upload training images

• Choose training zip files from /Chatbot-training folder, then choose ALL of them according to the class names including the negative sample

  

• Name each of the classes

  

• Copy the classifier ID

  

• Paste classifier ID in /Chatbot-docker/docker/Dockerfile

  

Add other Watson services

• Go to Bluemix catalog, search Conversation

  

• Name service as you want, and make sure there is a Credential name available then click on Create button

  

• Then you will see this page, and click on Service credentials, then copy username and password

  

• Paste username and password in /Chatbot-docker/docker/Dockerfile

  

• Start creating a new Conversation, click on Manage menu on the left then click on Launch tool

  

• For the first time visit of this tool, the IBM ID is needed for authentication, but as long as you've authenticated by Bluemix, it will treat you as a authorized user after clicking on the Log in with IBM ID by default. Otherwise you can try to sign in again with your Bluemix credentials

  

• Click on Upload icon to upload the training file, to locate the training file, go to /Chatbot-docker/trainings/conversation folder, find innovation-day.json file

  

• There will be one more panel named iChat, click on the menu on the top right, then select View details in order to copy workspace ID

  

• Paste workspace ID in /Chatbot-docker/docker/Dockerfile

  

• Go to Bluemix catalog, search Speech to Text

  

• Name service as you want, and make sure there is a Credential name available then click on Create button

  

• Then you will see this page below, go to next step refer to following step

  

• Copy username and password, paste them to Dockerfile and save the file

  

• Go to Bluemix catalog, search Text to Speech

  

• Then you will see this page below, go to next step refer to following step, then copy username and password

  

• Paste username and password in /Chatbot-docker/docker/Dockerfile

  

Update /Chatbot-docker/docker/Dockerfile with your Blockchain service UR, replace it with actual one if you're hosting the Blockchain service on the Bluemix (refer to Blockchain practice)

ENV APPLICATION_API_URL https://your_block_chain_service.mybluemix.net

If you are hosting the Blockchain service locally, obtain IP address first:

• macOS: Go to System Preferences -> Network, then copy your IP address

  

• Windows: Refer this guide to get the IP address

• Replace the URL with service URL (with your IP address)

  ENV APPLICATION_API_URL http://xxx.xxx.xxx.xxx:3000

Now you're ready for building the Docker image.

After installing the Docker, open Terminal on macOS or cmd on Windows

Build Docker image

• Go to /Chatbot-docker/docker/ folder under the Chatbot project root folder from the Terminal or cmd

• Find your Terminal from macOS

  

• Find your cmd from Windows

  

• Find your project folder, e.g. /path/to/Chatbot-docker/docker, then run the command:

  cd /path/to/Chatbot-docker/docker

• To build the Docker image, run the command by tagging it as chatbot:

  docker build -t chatbot -f Dockerfile ./

Run the Docker image on a new container

• Run the command:

  docker run -p 8888:9080 chatbot

Launch

Now you can visit your Chatbot web application via

• macOS: http://localhost:8888/Chatbot
• Windows: Refer to section Install docker on Windows 10 (5, 6), so the format of the URL would be like http://xxx.xxx.x.x:8888/Chatbot

Possible issues

• If you encounter an issue like the output from Terminal or cmd as below after running the docker build command:

   Sending build context to Docker daemon  4.246MB
   Step 1/15 : FROM websphere-liberty:webProfile7
  
   Get https://registry-1.docker.io/v2/library/websphere-liberty/manifests/webProfile7: unauthorized: incorrect username or password
  

• Please run this command:

  docker logout

Then try to re-build the Docker.