Create Amazon Alexa skill that uses IBM Bluemix OpenWhisk to host the backend actions and IBM Watson Tone Analyzer analyse an array of objects. With this in place, the user can play quotes that are happy, sad, fear ect. A user can ask things like:
Alexa ask watson to play a random
Alexa ask watson to play a happy quote
Alexa ask watson for a happy quote
If a happy quote is requested then Watson Tone Analyzer is used to find and play a quote where it has highest happy/Joy score.
- Amazon Developer Account
- Alexa Device (Optional)
- IBM Bluemix account. Sign up for Bluemix, or use an existing account.
Follow the steps below to create an Alexa skill, create the backend using IBM Bluemix OpenWhisk and use the Watson Tone to find a quote based on the user request.
-
Create A Skill
Login to the Amazon developer console, go to Skills -> Alexa Skill Kit (Get Started) -> Add a new skill. -
Give the skill a name:
Alexa & Watson -
Give the skill an invocation name:
watson
Invocation Name is what a user would speak into the Alexa device, for example I would say something likeAlexa ask watson to play a happy quote -
Next, under the Interaction Model, add the following:
-> Intent Schema{ "intents": [ { "intent": "randomQuote" }, { "intent": "happyQuote" } ] }Intent Schema is the actions that the user can preform, in plain text it's what Alexa can do
-> Slots
In this example we wont be using Slots.Slots are ways inject more specific user input into the Utterances. An Example where you would use Slots where if you ask Alexa: Find me a restaurants in Dublin area, the city in that case would be the Slot.
-> Sample Utterances
randomQuote watson to play a random quote randomQuote watson for a random quote happyQuote watson to play a happy quote happyQuote watson to for a happy quoteNote: sample Utterances is a way to map words to the intents.
Next, under the Configuration, we will call the server side action api to link the skill to the backend. For that we will be using IBM Bluemix OpenWhisk to create the Action then expose the action as an API.
-
Clone or fork the repository
-
Check out the code
-
Ensure your OpenWhisk command line interface is properly configured with:
wsk listNote: this shows the packages, actions, triggers and rules currently deployed in your OpenWhisk namespace.
-
Create a Watson Tone Analyzer service using:
cf create-service tone_analyzer standard alexaAnalyzer -
Obtain your Tone Analyzer credentials, from your Bluemix console under the service created you should be able to get the Service credentials.
-
Modify the action.js file by adding your Tone Analyzer username and password.
-
Create OpenWhisk Action
wsk action create alexaAndWatson action.js -
Expose your OpenWhisk Action as an API using:
wsk api-experimental wsk api-experimental create /alexaWatson /api/v1 post alexaAndWatson --annotation web-export true -
Copy the API endpoint in your console and past into the Amazon Alexa API filed, see image below. You should see something like:
ok: created API /alexaWatson/api/v1 POST for action /_/alexaAndWatson https://b4651037-d05a-4b7d-ab46-1e9a1abc096f-gws.api-gw.mybluemix.net/alexaWatson/api/v1
Next, under the SSL Certificate, select the My development endpoint is a sub-domain of a domain that has a wildcard certificate from a certificate authority option.
Test the skill by adding this text to the "Enter Utterance" text filed area:
Alexa ask watson to play a random
Alexa ask watson to play a happy quote
Output: "Here is a happy Quote that I found for you: happy wife means happy life".
Done...
- IBM Bluemix OpenWhisk site
- IBM Bluemix OpenWhisk keynote
- OpenWhisk Form Processing
- More IBM Bluemix OpenWhisk Samples
{
"document_tone": {
"tone_categories": [
{
"category_id": "emotion_tone",
"category_name": "Emotion Tone",
"tones": [
{
"score": 0.881688,
"tone_id": "anger",
"tone_name": "Anger"
},
{
"score": 0.022428,
"tone_id": "disgust",
"tone_name": "Disgust"
},
{
"score": 0.074133,
"tone_id": "fear",
"tone_name": "Fear"
},
{
"score": 0.006716,
"tone_id": "joy",
"tone_name": "Joy"
},
{
"score": 0.092043,
"tone_id": "sadness",
"tone_name": "Sadness"
}
]
},
{
"category_id": "language_tone",
"category_name": "Language Tone",
"tones": [
{
"score": 0,
"tone_id": "analytical",
"tone_name": "Analytical"
},
{
"score": 0,
"tone_id": "confident",
"tone_name": "Confident"
},
{
"score": 0,
"tone_id": "tentative",
"tone_name": "Tentative"
}
]
},
{
"category_id": "social_tone",
"category_name": "Social Tone",
"tones": [
{
"score": 0.022751,
"tone_id": "openness_big5",
"tone_name": "Openness"
},
{
"score": 0.234966,
"tone_id": "conscientiousness_big5",
"tone_name": "Conscientiousness"
},
{
"score": 0.366922,
"tone_id": "extraversion_big5",
"tone_name": "Extraversion"
},
{
"score": 0.567783,
"tone_id": "agreeableness_big5",
"tone_name": "Agreeableness"
},
{
"score": 0.001604,
"tone_id": "emotional_range_big5",
"tone_name": "Emotional Range"
}
]
}
]
}
}
| Commands | Description |
|---|---|
| $ wsk --help | Wsk help to find available commands |
| $ wsk activation poll | Useful for troubleshooting to view logs and errors |
| $ wsk action create 'Action_Name' 'Action_File' | To create OpenWhisk action |
| $ wsk action list | list all OpenWhisk actions |
| $ wsk api-experimental create /'Api_Name' /'Api_Path' post 'Action_Name' | Expose action as a API |
| $ wsk api-experimental list | List all API's |
See License.txt for license information.
This project is a sample application created for the purpose of demonstrating a serverless app with OpenWhisk. The program is provided as-is with no warranties of any kind, express or implied.