Concepts

Workspaces

Workspaces are logical containers that host your Conversational AI bot/VA .Morfeus supports multi-tenancy and you can create multiple bots with a logical separation with different access controls for all bot related configuration including data, workflows and fulfillments, settings and customer data.

Intents

Intents are the basic block of how conversations are identified and classified. User's query is initially classified into the following

• Query - User tries to enquire about a personal information. Example: "what is my balance"
• Transaction - User is trying to perform an action. Example: "send 50 to amol"
• FAQ - User is requesting from a knowledge base. Example: "What is neft?"
• Smalltalk - General chit chat to bootstrap the conversation. Example "good morning"

qry-balanceenquiry txn-moneymovement txn-blockcard

Entities

Entities are mechanism for identifying and extracting useful data from natural language inputs.

While intents allow understanding the motivation behind a particular user input, entities are used to pick out specific pieces of information that your users mention — anything from person names to product names or amounts with units. Any important data you want to get from a user's request will have a corresponding entity.

EX: EntityName - sys.city

Dialogs

Queries or transaction intent can have more than one dialog turn to complete the user's request. These type of intent can have multiple dialog turns. One dialog turn involves providing the following three information

• Expected entity
• Bot says
• User says

SmallTalk

These are polite conversations about unimportant or uncontroversial matters, especially as engaged in on social occasions. These help to interact with users informally before getting into real conversations.

EX : Hi, How are you , etc

CognitiveQnA (FAQ)

The Cognitive QnA service answers your users' natural language questions by matching it with the best possible answer from the semi-structured content such as Frequently Asked Question (FAQ) in your Knowledge base. The easy-to-use admin interface enables you to create, manage, train the service without any development experience.

Keyphrases

Keyphrases are significant phrases in a FAQ like a product e.g credit card or an action e.g apply. CognitiveQnA uses keyphrases to improve the accuracy of responses and tags them automatically during the training phase.

SpellCorrection

Spellcorrectors helps Grammatically to eliminate errors and find the perfect words to Express user utterances.

Example:

i have=i've i would=i'would i am=i'm i will=i'll

Acronyms

Acronyms are custom ways to provide alias to certain words to improve the vocabulary of the bot.

Example:

transfer = tsfr credit card = cc, credit, cr. card platinum edge card = pe card, pec, platinum

NLU (Natural Language Understanding)

It enables a deeper understanding of the structure and meaning of a user utterance. Attributes of a sentence like discourse relationship, response type and grain, tense and type can be extracted and used to fine-tune conversation responses to users.

Context Expansion and Context Change

Humans and contextual in their behaviour , Context expansion is able to expand follow on partial queries by deriving key attributes of the parent conversation. For e.g what is the balance on my visa card , followed by , pay the min due. Is expanded to pay the min due for my visa card.

Context change is ability to detect when users are deviating from a logical conversation chain e.g if a user is switching from paying his bill to applying for a new card.

Knowledge Graph

A knowledge graph is a hierarchical mapping of the products and services, an enterprise has to offer for customers. AI Classifiers are probabilistic by nature, so they may not be best suited for short 2-3 word utterances. Knowledge Graph provides a deterministic alternative with configurable probing in case of ambiguity introduced when processing the customer provided utterances.

For example, if the customer wants to know about a product launched newly in the market, so the customer may ask "Active debit card", so AI will face ambiguity in making decisions processing these utterances.

In these cases, the Knowledge graph will let the business users configure their products and services (Active AI Credit card) with relevant suggestive actions (Apply), which will drive the customers.

Functionality

Logical grouping of training data, fullfillment, templates, and messages for a particular use case.

Conversational Workflows

It is a low code framework for designing and fullfilling a conversational journey, enables developers to design conversational journeys via a WYSWYG web based Workflow Editor. Developers can specify slots (these can be provided in multiple sequences in utterances) , format responses and messages , script validation , configure routing , invoke API's and much more.

Channels

Out of the box connectors to standard Messaging Apps , IOT devices , Voice , Mobile and Web SDK's.

Templates

Out of the box generic UX components like Lists, Carousel, etc which can be edited by a WYSWYG web-based Template Editor. Templates are rendered automatically across multiple conversational channels and can be scripted or customized on compatible channels.