Cognitive Model

The cognitive model in Anthology Digital Assistant is a hierarchical design that allows for future investment into specific areas of the Digital Assistant's intelligence. Each cognitive model is made up of an independent LUIS trained model. Each model will be licensed independently so that each customer can have access to different intelligence sets, or skills, or other LUIS models.

Anthology Digital Assistant includes the following cognitive models:

  • Dispatcher
  • General
  • ChitChat
  • FAQ

Renee cognitive model structure

Dispatcher

Since the Digital Assistant incorporates more than one model, the Dispatcher is needed to determine which LUIS model or QnA Maker knowledge base best matches the user input. The Dispatcher does this by identifying the intent of the conversation and creating a single LUIS parent app to route user input (utterances) to the correct model.

General

The General model contains intent definitions such as greetings and handles commands to start, stop, repeat, escalate, etc. a conversation.

ChitChat

The Digital Assistant's personality was seeded from the professional ChitChat model that comes with the Microsoft Azure Bot Framework. We have taken this model and refactored it in the structure below so that we can focus on specific aspects of the Digital Assistant's personality within the code but also escalate issues to advisors at the university when students show signs of inappropriate behavior or show signs of really needing to connect with someone. The goal is to be able to organize the code within the LUIS model to make it easier to refine each of these areas and train and release new models as more people interact with the Digital Assistant's general intelligence and personality.

Chitchat model design

For more details about the ChitChat model, see https://github.com/microsoft/botbuilder-tools/tree/master/packages/Ludown/ChitChat.

FAQ

Anthology Digital Assistant has the ability to provide institution-specific answers to questions about Admissions, Academics and Campus Life, Advising, and Financial Aid. Within each area, the Digital Assistant also captures questions that should raise flags for follow up from qualified members of the institution. Any questions trained in these areas will attempt to transfer to an available agent or open a case in Anthology Reach on behalf of the user and log the conversation.

FAQ for higher Education

To review the default question and answer pairs provided in Anthology Digital Assistant, see FAQ Model - Sorted by ID and FAQ Model - Sorted by Category.

Student Verification

The Student Verification model was developed with the idea that non-Anthology Reach customers will be able to access Anthology Digital Assistant. In this case, standalone-Student Verification customers will be able to interact with the Digital Assistant but only have access to ChitChat and Student Verification.

This allows for two scenarios:

  1. A customer using Anthology Reach and Student Verification would have full control over all of the answers Anthology Digital Assistant provides.

  2. A stand-alone Student Verification customer will access the Digital Assistant through a JavaScript chat client. Unauthenticated users will have access to a set of default answers to financial aid questions. In this case, the Digital Assistant will not write to the Contact's timeline or escalate issues to a person at the institution. Anthology Digital Assistant would be more like a knowledge base.

The Student Verification model is available in Anthology Digital Assistant version 2.0.0. The LUIS model for Student Verification answers questions related to the verification process for students and parents who use that product.

For details about the questions and answers in the Student Verification model, see FAQ Model for Student Verification - Sorted by ID and FAQ Model for Student Verification - Sorted by Category.

Anthology Reach Contact Update

Future releases of Digital Assistant will include skills that can be added to actively interact with other applications. The Anthology Reach Contact Update cognitive model will allow users to directly interact with their demographic data and update their contact information, make appointments with advisors, and register for events. This model has not yet been developed.

Anthology Student Registration Assistant

Future releases of Anthology Digital Assistant will include skills that can be added to actively interact with other applications. The Anthology Student Registration Assistant will interact with a student's academic record including Degree Audit, Degree Pathway, and Course Registration. This model has not yet been developed.

The cognitive model in Anthology Digital Assistant is a hierarchical design that allows for future investment into specific areas of the Digital Assistant's intelligence. Each cognitive model is made up of an independent LUIS trained model. Each model will be licensed independently so that each customer can have access to different intelligence sets, or skills, or other LUIS models.

Anthology Digital Assistant includes the following cognitive models:

  • Dispatcher
  • General
  • ChitChat
  • FAQ

Renee cognitive model structure

Dispatcher

Since the Digital Assistant incorporates more than one model, the Dispatcher is needed to determine which LUIS model or QnA Maker knowledge base best matches the user input. The Dispatcher does this by identifying the intent of the conversation and creating a single LUIS parent app to route user input (utterances) to the correct model.

General

The General model contains intent definitions such as greetings and handles commands to start, stop, repeat, escalate, etc. a conversation.

ChitChat

The Digital Assistant's personality was seeded from the professional ChitChat model that comes with the Microsoft Azure Bot Framework. We have taken this model and refactored it in the structure below so that we can focus on specific aspects of the Digital Assistant's personality within the code but also escalate issues to advisors at the university when students show signs of inappropriate behavior or show signs of really needing to connect with someone. The goal is to be able to organize the code within the LUIS model to make it easier to refine each of these areas and train and release new models as more people interact with the Digital Assistant's general intelligence and personality.

Chitchat model design

For more details about the ChitChat model, see https://github.com/microsoft/botbuilder-tools/tree/master/packages/Ludown/ChitChat.

FAQ

Anthology Digital Assistant has the ability to provide institution-specific answers to questions about Admissions, Academics and Campus Life, Advising, and Financial Aid. Within each area, the Digital Assistant also captures questions that should raise flags for follow up from qualified members of the institution. Any questions trained in these areas will attempt to transfer to an available agent or open a case in Anthology Reach on behalf of the user and log the conversation.

FAQ for higher Education

To review the default question and answer pairs provided in Anthology Digital Assistant, see FAQ Model - Sorted by ID.

Student Verification

The Student Verification model was developed with the idea that in future releases non-Anthology Reach customers will be able to access Anthology Digital Assistant. In this case, standalone-Student Verification customers will be able to interact with the Digital Assistant but only have access to ChitChat and Student Verification. The Student Verification model is an extract of the Financial Aid section of the FAQ Model - Sorted by ID.

This allows for two scenarios:

  1. A customer using Anthology Reach and Student Verification would have full control over all of the answers Anthology Digital Assistant provides.

  2. A stand-alone Student Verification customer will access the Digital Assistant through a JavaScript chat client. Unauthenticated users will have access to a set of default answers to financial aid questions. In this case, the Digital Assistant will not write to the Contact's timeline or escalate issues to a person at the institution. Anthology Digital Assistant would be more like a knowledge base.

This model will not be available with Anthology Digital Assistant version 1.x.

Anthology Reach Contact Update

Future releases of Digital Assistant will include skills that can be added to actively interact with other applications. The Anthology Reach Contact Update cognitive model will allow users to directly interact with their demographic data and update their contact information, make appointments with advisors, and register for events. This model has not yet been developed.

Anthology Student Registration Assistant

Future releases of Anthology Digital Assistant will include skills that can be added to actively interact with other applications. The Anthology Student Registration Assistant will interact with a student's academic record including Degree Audit, Degree Pathway, and Course Registration. This model has not yet been developed.