Clinc Knowledge-Base

# Clinc AI Platform Terminology

This document defines common terms used throughout the Clinc Knowledge-base.

A | B | C | D | E | I | J | M | N | O | P | Q | R | S | T | U | W

# A

# Artificial Intelligence (AI)

Artificial Intelligence is the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, natural langauge processing and machine translation.

# AI Project

An AI projects is the workplace to build your conversational AI experience. It packages all the versions and snapshots into one project for easier project management.

# AI Version

An AI version is a working branch of your AI project. It is the workplace to build competencies, collect and curate training data, train models (classification & SVP), and customize responses.

# Applications Programming Interface (API)

Application Programming Interface is a set of subroutine definitions, protocols, and tools for building application software.

# Automated Speech Recognition (ASR)

An engine enables conversational AI to process voice-driven interactions and communicate outside of traditional text-based interfaces.

# Amazon Web Services (AWS)

Amazon Web Services is a cloud computing platform provided by Amazon which provides a mix of infrastructure as a service (IaaS), platform as a service (PaaS) and packaged software as a service (SaaS) offerings.

# B

# Business Logic

Code that interfaces with a database or Application Programming Interface (API) outside the Clinc AI Platform. It allows you to execute any code/logic you want when a user makes a certain query.

# Business Logic Server

Server that allows you to expose an endpoint for the Clinc AI Platform to POST request a JSON payload to. It hosts your business logic codes and can verify slots or change state,it can also execute actions and get data to pass back to Clinc Response generation.

# Business Logic Transition

Business Logic transition is a whitelisted transition that can be arbitrarily initiated by Business Logic. If Business Logic overrides the destination state, and a matching Business Logic transition exists, then the transition will be taken.

# Bot Orchestrator

A bot orchestrator is a network or "mesh" of models that are connected through transitions and act as a singular AI version.

# Bot Orchestrator Transition

Bot orchestrator transitions describe how to transition from orchestration socket to another. They are used to determine what steps the bot orchestrator should proceed with. There are three types of bot orchestrator transitions, classification transitions, variable transitions, and out of scope transitions.

# C

# Channel

A channel is devices like Alexa, Google home, or mobile (android and IOS), web, IVR and Facebook Messenger to which you can launch your virtual assistant for end users to interact with.

# Checkout

Checking out a snapshot lets you create a working branch (version) to build conversational AI experience.

# Class

A class is the label that defines the direction of the next turn in the conversation. See also intent.

# Classification (CLF)

Classification is the process that determines the direction of the next turn in the conversation.

# Classifier

A classifier/query classifier is an AI engine that infers the direction of the next turn in the conversation.

# Classifier Intent Class

Intent classes are used to train the classifier and is a superset of the training dataset of the source socket state’s native classifier, augmented with data curated at the Orchestration level. In order to allow classification to occur on Orchestration source socket states to trigger CLF mesh transitions, the "Orchestration Intents" or triggers on the mesh transition must be trained into the classifier as intent classes.

# Classification Job

A type of crowdsourcing job that engages crowdsource workers to provide utterances, things that people say, about a specific intent.

# Classification Transition

Classification transition is a classification-driven transition. When processing a query, the Clinc AI engine begins with classifying the utterance. After classification, the AI engine looks at the outgoing classification transitions from the current state. If an outgoing transition includes the resulting class, the transition can be executed successfully otherwise the conversation is taken back to the root state.

# Competency

A competency can be thought of as a topic area or an atomic action that your AI “brain” knows about or can carry out. For example, asking about the weather, ordering food, or setting cruise control speed etc.

# Conversational State

See State.

# Crowdsourcing

Crowdsourcing is a way to quickly collect diverse data/utterances for a competency by leveraging many sources from inside or outside your organization.

# D

# Data

In the Clinc AI Platform, we gather utterances as training data. Utterance is a question or statement that a user may ask the AI in order to get a response about the selected topic.

# Data Query Language (DQL)

Data Query Language (DQL) is a language that Clinc has developed that enables you to locate data through queries. It supports queries based on strings, slots, slot values, phrases and the number of slots labeled.

# Deep Neural Networks (DNN)

A Deep Neural Network is an Artificial Neural Network (ANN) with multiple hidden layers between the input and output layers. DNNs can model complex non-linear relationships.

# Deployed Version

A deployed version is a snapshot version assigned to an application inside a production environment. It will be prioritized in the cache for querying.

# Destination Socket

Also known as an ingress socket. This term is used with the bot orchestrator when describing the competency/state in which orchestration transition directs the conversations state to .

# I

# Integration Suite

An integration suite is a grouping of integrations tied to a single ai version. This can be visualized as creating a "virtual assistant" that can be deployed to any number of supported integration platforms.

# Intent

In the Clinc AI Platform, an intent refers to a collection of data that is associated with classification transitions. Intents are used to train the query classifier AI engine so that the AI engine can decide what direction the conversation is going to take.

# J

# JavaScript Object Notation (JSON)

JSON is an open-standard file format that uses human-readable text to transmit data objects consisting of attribute-value pairs and array data types.

# Jinja

A powerful templating language used to translate response templates into final natural language by filling in variables to construct the final response.

# M

# Machine Learning (ML)

Machine Learning is an application of Artificial Intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. Machine learning focuses on the development of computer programs that can access data and use it learn for themselves.

# Mesh

See bot orchestrator. Also known as, Orchestration Mesh, Orchestration, Bot Orchestrator

# MTurk

The platform provided by Amazon that is integrated with the Clinc AI Platform for users to crowdsource data to train their conversational AI.

# N

# Natural Language Generation (NLG)

As a subset of NLP, NLG transforms structured data into natural language. Generally, NLU takes up the undertanding of the databased on grammar, the context in which it was said and decide on intent and entities, NLP converts a text into structured data and finally, NLG generates a text based on structured data.

# Natural Language Processing (NLP)

Natural Language Processing is the ability of a computer program to understand human language as it is spoken. NLP is a component of Artificial Intelligence (AI).

# Natural Language Understanding (NLU)

The field of NLU is an important and challenging subset of natural language processing (NLP). Understanding of natural human languages enables computers to understand commands without the formalized syntax of computer languages and for computers to communicate back to humans in their own languages.

# O

# Orchestration Graph

The orchestration graph is the collection of sockets and transitions displayed on the Clinc AI Platform. It serves are the building site for orchestrations. If a mesh exists, it can be viewed and manipulated on the orchestration graph.

# Orchestration Intent

An orchestration trigger refers to a collection of data that is associated with orchestration classification transitions. Synonymous with orchestration trigger.

# Orchestration Root Socket

An orchestration's root socket is the entry point for the orchestration mesh.

# Orchestration Trigger

See orchestration intent.

# Orchestration Utterance

A single sentence or phrase used in a orchestration intent data set. Orchestration trigger data sets consist of a whole collection of orchestration utterances.

# P

# Personality

A set of response templates in different tones of voice across a set of competencies which you can deploy onto a variety of devices.

# Pre-built Slot

Pre-built slots are platform built-in slot extractors to help you quickly extract values that belong to that specific slot. It doesn't require training data upfront and is non-configurable.

# Q

# Query

A query is any request given by the user to a conversational AI.

# Query Sidebar

Also known as QSB. The query sidebar provides Clinc Platform users with a convienent way to query

# R

# Reclassify

When this flag is set, a user can force the AI Version or Bot Orchestration to reclassify, in other words, have the model process the query again. The reclassify flag will cause the query to be classified again using the classifier located at the transition's destination state. A user can edit state settings to have some control over the behavior of the reclassification.

# Recurrent Neural Network (RNN)

A Recurrent Neural Network is a type of advanced Artificial Neural Network (ANN) that involves directed cycles in memory. They are typically applied to sequential data such as speech.

# Response Generation

The process in which the AI uses response templates and information identified from the user query or from business logic to generate a response.

# Response Template

Structure of how a response is built in natural language with slots/variables that are dynamically assigned at response time.

# Retain State

Retain state will keep the context of the conversation by keeping the slot values throughout each transition.

# S

# Scoping Job

Scoping job is a type of crowdsourcing job available on the Platform to help you gather ideas about your AI version. It is useful when you have a vague idea what you want to build, but aren't exactly sure what users would want or how they would use it. Scoping data offers a way to get a sense for end-users' perspective on a potential product functionality.

# Secondary Cluster

In slot mapping, second cluster values are what you configure to have the svp extracted slot map to. They are usually synonyms or closely related terms to the target value.

# Slot

Important information/variables in an utterance.

# Slot Labeling Job

A type of crowdsourcing job that engages crowdsource to label slot data to train the SVP engine. A slot labeled needs to pass a certain number of votes to count.

# Slot Mapping

Slot mapping is the process of mapping an identified slot value (word/phrase being extracted from the user query, also called SVP value in the following text) to a set of secondary clusters (word/phrase) through a mapping type.

# Slot/Variable Transition

Slot transition is a slot value-driven transition. This type of transition requires a set of slots to be extracted. When all the slots are filled, a transition will automatically be executed. Until then, the AI prompts the user for the information needed to make the transition.

# Slot-Value Pairing (SVP)

An AI engine that extracts values from an utterance and pairs them to the appropriate slot variable.

# Snapshot

A Snapshot is a non-editable version. It serves as milestones of the project from which team members can diverge as well as a fallback to revert to in the event of unintended changes or human errors.

# Socket

A bot orchestration socket is a representation of an AI Version on the bot orchestrator. A socket can have many bot orchestration transitions. The root socket on a bot orchestration graph is the entry point for the mesh.

# Source Socket

Also known as an egress socket. This term is used when describing which competency/state an orchestration transition is coming from.

# State

A state is any point in the conversation where your AI might want to generate a response.

# State Graph

The State Graph is a visual representation of all the states and transitions inside of the competencies of your conversational AI.

# T

# Test Assertion

In the context of Clinc AI platform, a test assertion is to check if a part of the AI pipeline (intent, slot, slot mapping or response) matches expected intent. A quick way to assert expected intents, slots, slot mappers and response values is through Autofill. You can also choose to manually assert when the value doesn't match your expectation.

# Test Case

A test case is an utterance/query to test the performance of classification, SVP, slot mapping or response separately or all at once. Test data should always be separated from training data.

# Test Dialog

A test dialog is a set of test cases that are intended to test the flow of a conversation.

# Terminal State

Defines a state that has no outgoing transitions. Competencies are only considered to be completed after the user has reached a terminal state. After the conversation has reached a terminal state, the conversation returns to the root state.

# Test Suite

A test suite is a collection of full conversations (end-to-end testing) and unit tests (classification unit testing, svp unit testing etc.).

# Tokenization

The process of segmenting running text into words and sentences.

# Transition

A transition is a way through which the conversation can move into the next state.

# U

# Utterance

A question or statement that a user may ask the AI in order to get a response about the selected topic.

# W

# Webhook

The mechanism to expose Business Logic using Hyper Text Transfer Protocol (HTTP) in Clinc’s AI Platform.



Last updated: 08/21/2020

← AI Projects →