Knowledge Base User Guide
Knowledgebase is an add-on to the chat solution and provides agent assistance based on the articles added in the system. Authorized users add articles using the Knowledgebase Author. KB Author is a web-based user interface for adding knowledge base articles in the system.
Role-based access based on UMM
An authorized user should have the trainer role to use KB Author. See UMM documentation about adding users and roles in UMM.
Articles
Click on “Add New” to add a new article. An article has the following three fields:
Field | Data type | Description |
Intent | string(30) | An ‘Intent’ is a label of the question. It should be a distinguishable short phrase. For example, for a question “What are the office timings” the Intent could be “office_timings” An Intent can be of a maximum of 30 characters. |
Question | string(1000) | A question is like the subject/title of an article. Add multiple synonym questions for a better prediction of it’s matching with the customer utterance. Add at least 2 questions per article |
Answer | string(2000) | Describe the answer in plain text |
An authorized user can add, update, delete articles from the system.
Edit an article
Select any article from the list
Article opens in edit mode
Make required changes
Click on update button to save changes
Delete an article
Select any article from the list
Article opens in edit mode
Click on delete button.
Alert will appear to confirm before delete
Click on yes to confirm
Article is deleted and removed from the list.
Training
Articles added in the system are not available to the agent until an Author trains the system. For training set of articles for a particular language,
Select the language (agent) from the dropdown.
Select Train from the detail submenu.
Search an article
An author can search an article from the search interface based on:
Intent name
Text in first question of the article
Logout
To logout from the KB Author interface, select Logout from the menu. See the screenshot below. All the articles added in the system will persist but articles will be passed as agent suggestions only after training.
