Day 3 | Skybuffer Community Chatbot | How to Create Multilingual Question-and-Answers Scenario Fast and Easy
Question-and-Answers (QNA) skills are an integral part of any business content portfolio. Skybuffer offering provides simple mechanisms of creation and configuration of your multilingual QNA skills out-of-the-box based on Skybuffer open-source community AI content.
Please, follow these simple steps to create your first QNA scenario with multilingual support out-of-the-box:
Step 1: Find zxas-template-fallback skill in your forked version of Skybuffer Foundation Content chatbot:
Step 2: Press Fork for the selected skill:
Step 3: Select the chatbot where the skill is to be forked:
Step 4: Press Fork and wait for the Success message to appear in the bottom left-hand corner of the screen:
Step 5: Go to the Build tab. Use Add skill group function to create a group for your QNA scenario:
Step 6: Input the name of the group and press Create Group:
NOTE: Names of the skill group and skills themselves should start with the registered Customer namespace that you have registered in Skybuffer Customer Namespace application.
Step 7: Find your forked skills and get into the skill by clicking on it:
Step 8: Switch to the Edit mode and rename the skills:
Step 9: Hit ENTER and make sure that the name has changed:
Step 10: Repeat steps 8-9 for the forked business fallback skill.
Step 11: Return to the Build (skills list overview) tab, mark your new skills:
Step 12: Add the selected skills to the skill group:
Step 13: Create Intent for your new QnA skill. Go to the Train tab. Create new Intent using the same name as for skill-trigger of your new scenario:
Step 14: Enter description of the intent and hit CREATE INTENT:
Step 15: Go to your brand-new Intent and fill it out with phrases that are relevant for triggering of the new QnA skill. General SAP recommendation is to input not fewer than 50 phrases. You can use manual input, import from file and suggestions enrichment functionalities.
NOTE: Phrases should be inputted in English only.
Step 16: Once you are done with training, press the Train button to enrich NLP model with new training data:
Step 17: Start adaptation of the skills. Go to the Build tab and select the trigger of the new skill. Replace the standard description of the template skill with your scenario-specific description in README.md section. Press Save:
Step 18: Go to the Trigger tab and replace the template intent with your brand-new skill intent:
Step 19: Press Save. Now your Trigger condition is ready and should look as follows:
Step 20: Go to the Actions tab. Switch to the Edit mode in the first logical block:
Step 21: Replace the value of parameter rt_return_to_function with the name of your trigger skill:
Step 22: Press Save.
NOTE: This value should be used for categorization configuration (please, refer to Day 2 blog post).
Step 23: Replace the text of [Your reply in English] in two logical blocks of the skill:
English block: update the text section.
Switch to the Edit mode, input the text of the reply. Enable Markdown syntax if relevant. Press Save:
Non-English block: update rt_source memory parameter value.
Input the same text instead of [Your reply in English]:
Step 24: Modification of the business fallback skill. Access your new business fallback skill, replace README.md text:
Step 25: Access the Trigger section:
Step 26: Replace the value of the memory parameter _memory.rt_intent.slug with the name of your new intent without @ sign. Hit Enter:
Step 27: Your fallback skill adjustment is now completed.
Step 28: Test your new skill ready to reply technically in any language (replies management steps are following):
Step 29: Customize your bot replies translation using the Chatbot Vocabulary application (find more details in Day 2 blog post )
Go to https://discover.skybuffer.com/ to access the Bot Management Apps.
NB! you can use administer user ID for our discovery cloud organization:
Discovery User: HC_DEMO
NB! In case you do not capture the translation in the vocabulary, /translate web hook can always give you Google translate API key to translate your reply from rt_source in the runtime.
In the Chatbot Vocabulary application you should provide the skill ID, the target language and the phrase:
NB! Please use exactly the same phrase you have placed into the rt_source memory parameters and keep JSON formation that is set by SAP Conversational AI (quotes in our case).
After you save it, the chatbot will take the phrase from the Vocabulary.
Now you have completed Day 3 guidelines and you know how to build a support chatbot that can handle QnA tasks in any language. Roughly it takes you only 5-10 minutes to add a QnA skill to the chatbot.
Generally speaking, after you go through Day 1, Day 2 and Day 3 and follow the guidelines, your chatbot will be able to:
- Capture verified users’ contact details and generate new leads for you
- Speak about the provided services
- Seamlessly integrate an operator
- Categorize conversations
- Provide replies in various languages without any additional training
- Provide customized replies in various languages that are not translated automatically
- Capture support requests in case operators are offline or Hybrid Chats are connected to SAP Conversational AI chatbot in the operator-free mode
- Save all conversations so that you could always review them
- Provide information according to QnA knowledge base that is added as a set of QnA skills
Looks like it is ready to Go Live and support your Clients, Employees or Business Partners, what would you say?
P.S. You can also find the entire list of our blog posts under the links below: