Answers v2

Document Title: Answers v2beta

Answers v2 Version: V2

Creation Date: Aug 31, 2021

Author: ORBITA

 

We at Orbita thrive to give the best experience to our customers. One step in that direction is the launch of a new beta version of the Answers v2 interface which is available after the v4.0.90 upgrade.

The Answers v2 has a renewed, enriched, and fresh user experience, designed to help you focus on what is relevant. The new interface allows you to create and manage your knowledge bases and the corresponding questions with ease.

 

What does the beta label mean? Orbita Answers vNext beta launched in version 4.0.90 in September 2021 with certain known limitations. These limitations are:

  • Migration Support - You cannot migrate Answers v1 Knowledge Bases to Answers v2. Migration support is in progress and will be available when the beta label is removed from the product. If you have an Answers v1 Knowledge Base that you would like to have migrated to Answers v2, please reach out to your Customer Success account manager to discuss. UPDATE: Migration support will be available in V.0.98

  • Custom Schemas - Answers v2 supports knowledge bases leveraging custom schemas from v4.0.94. UPDATE: Migration support will be available in V.0.98

  • Documentation - While there is documentation available to support Answers vNext beta , forthcoming documentation will be provided to support migration and configuration options.

 

User experience simplified

We understood that it’s difficult to get started with Orbita Answers and it often takes organizations many weeks to implement a question/answer experience using Orbita Answers because of the steep learning curve. It requires knowledge of the knowledge base, content structures, architecture, etc to effectively use Answers. To address these issues we have simplified the user experience to place a greater emphasis on usability without getting into too many details. The Answers author will see only the content that is relevant to create an effective Knowledge base and have the ability to organize the content in a tree view for better comprehension.

Answers v2 landing page

You can create a new KB using the “Create a Knowledge Base“ button. All the KBs will be listed here.

Header Title

Description

Knowledge Base

The name of the Knowledge Base

Type

The content schema used while creating the KB.

Channels

The Channels such as text, voice, and screen in which the content can be handed out.

Primary Questions

Shows the primary question.

Answers

Shows type of answers content created ( Text, voice, screen/image)

Last Updated

Shows when the fact has updated

Creating new KB

Build tab

This tab lists all the facts you authored in a table format.

 

The first column in the table displays an arrow that is tilted up which shows if that question has children in the tree view.

Hover over the headers for a tooltip.

Header Title

Description

Header Title

Description

Primary Question

Shows the Primary question.

Labels

Lists all the labels created for each Primary question.

Variations

Shows the count of question variations created included.

Answers

Shows type of answers content created ( Text, voice, screen/image)

Last Updated

Shows when the fact has updated

Organize tab

The questions in the KB can be organized in a tree view in this tab.

There are two types of Root type options in the Answers v2 Manager node. They are Ancestor and Parent types. More information on Root types is here Answers v2 Manager node | Answers v2 Manager node properties

 

Advanced tab

This tab includes Settings, and Import, Export of the CSV files.

Refer How do I create a .xlsx file for Answers v2? for details on how the .csv file is structured.

Question & Answer Details page

This is where you define your Question, Answers, Labels, Question variations, and so on.

Primary question

Primary Question is the canonical question for each Question & Answer.

The primary question also sets the question context within the session until the next primary question is triggered.

Any ambiguous phrase such as “what are the symptoms?” that triggers a label intent and which is not a primary question will be linked to the question context (if present) such as “Cancer”. The corresponding answer (symptoms of cancer) is delivered to the user.

Each primary question will have a corresponding Intent created in the Intent library.
You can see the Intent ID created for a Primary question in the IDs section.

You can search the intent library for the same intent ID.

Label

Though not mandatory, it is highly recommended to add Labels for each Question & Answer.

Labels help Orbita disambiguate an ambiguous phrase like “what are the symptoms?” to an answer, given the user’s context (e.g. cancer)

Labels also serve as a way to categorize a Question & Answer if there is no prior context established.

Labels can be referenced in question variations. The syntax is to use the @ sign to reference them. Both labels & their synonyms can be referenced in the question variations.

Orbita creates one label intent and one label slot per KB. The Label slot is mapped to the Label intent for that KB.

Each label you create will add a new line item in the label slot.

Synonyms of a label are added to the respective slot line item as synonyms.

When there is a common Label given to a set of questions, the system cannot recognize the correct Primary question answer. The User must provide a primary question or its variations.

 

Synonyms

You can add synonyms to the Labels using the Synonyms button at the top right corner of the page while editing the contents of the KB.

On the Synonyms window, you can add/delete the synonyms for each Label. The changes will be reflected in the slot list.

Architecture

For example, consider the below scenarios where you have the questions as shown in the screenshot below.

  1. The chatbot user asks an unambiguous question “What is diabetes?” as the utterance is associated with the Primary question (or one of its variations) the question context is set to this primary question.
    Here, if the user utterance is “diabetes“ then the label intent will get triggered. Since the diabetes label is associated with only one primary question, the question context is set to this primary question.

    1. The chatbot user then asks “What are the symptoms?“. This will trigger the label intent and since the label “Symptoms” is associated with two different Primary questions, it is an ambiguous question.
      Here, since there is not a single answer to this question, the chatbot will ask the user to rephrase the question.

  2. The chatbot user asks a question “What is cancer?” which triggers a Parent primary question and sets the context.

    1. The chatbot user then asks “What are the symptoms?“, the bot will now understand that the chatbot user is asking about the “symptoms of cancer” and provides the respective answer.

 

Related Articles