Salesforce

The Veritone data model

« Go Back
Information
The Veritone data model
000004278
Public
Product Selection
aiWare - aiWare
Article Details

This diagram shows a high-level overview of the data model implemented in the Veritone GraphQL API.

 

Entity relationship diagram

Typically, each block in the diagram has a corresponding type in the GraphQL schema. Each type has corresponding queries that retrieve it and mutations to modify it.

This example shows the Task and Job blocks from the diagram schema in GraphQL:

type Job {
  # Retrieve all tasks contained in this job
  tasks: [Task]
}

type Task {
  # Retrieve the job that contains this task
  job: Job
}

type Query {
  # Retrieve a single job by ID
  job(id: ID!): Job
  # Retrieve a list of jobs
  jobs: [Job]
  # Retrieve a single task by ID
  task(id: ID!): Task
}
[Note] This schema is simplified with basic fields and no filtering or paging.

Type descriptions

This section describes what each entity type means.

EntityDescription
ApplicationA custom Veritone client application. Third-party developers in the Veritone ecosystem can create, deploy, and market their own applications using Veritone Developer.
AssetA piece of data. An asset can be consumed by an engine for analysis, or it can be produced by an engine to store results.
BuildA version of an engine. Engine developers upload an engine build to deploy or update their engine; a build represents the engine code. Each engine can have any number of builds representing different versions of the code, only one of which is active at any given time.
CollectionA group of related data. A collection can be placed in a folder, or not. Collections can be shared and publicized.
EngineA self-contained, encapsulated program that is integrated into the Veritone platform and performs either AI analysis or infrastructure functions such as ingestion. Third-party developers in the Veritone ecosystem can create, deploy, and market their own engines using Veritone Developer.
Engine CategoryA group of engines with similar features. Expected engine payload and output are common to engines within an engine category. Engine categories make it easier for users to find and select engines.
Engine Whitelist/BlacklistA list of users in an organization allowed to search, view, and use a set of engines available to the organization. These lists are managed by Veritone in accordance with the organization's needs.
EntityAn aggregation of assets, or entity identifiers, for a defined concept, such as a person, a company or organization, an advertising campaign, a type of object, etc.
Entity identifierAn asset associated with an entity. Examples are headshots for face recognition, voice clips for speaker recognition, ad creative for audio fingerprinting, DLM for transcription, and aliases for transcription.
Entity identifier typeThe type of asset associated with an entity. Examples include headshot, logo, voice clip, ad, and DLM. The library type defines what identifier types it can support.
FolderA container for organizing and sharing information, including temporal data objects (media and other data), watchlists, and other folders.
JobA job describes all the task necessary to perform a single high-level operation, such as run a series of engines against a piece of data or ingest data from a given source.
LibraryA named collection of entities an organization is interested in identifying in media, e.g., American politicians. A library's type defines what type of entities it can hold.
Library collaboratorAn external organization that a library has been shared with. Users in the collaborator organization can view the library and use it to train their own models, but cannot modify it.
Library engine modelA set of data generated during an trainable engine's training step. The model is (in some cases) provided to an engine when the engine is run. A model can optionally contain an asset representing the model data.
Library typeA description of the type of entities the library contains. Used to tailor the UI experience for specific types of entities (for example, People, Ads, etc).
MentionA record that a given entity, tag, or search condition was matched (or "mentioned") in an engine result. Mentions can be shared and published.
Mention commentA user comment on an individual mention. Comments are used to share, publicize, and collaborate on mentions.
Mention ratingA user rating on an individual mention. Ratings are used to share, publicize, and collaborate on mentions.
OrganizationA Veritone platform subscriber. An organization can represent a company, a department within a company, or an individual developer.
PermissionA single functional permission on the Veritone platform, such as "create library" or "view asset".
RoleAn abstract, business-focused way of describing a user's permissions on the Veritone platform. For example, a "Library Editor" manages libraries on behalf of the organization. A role collects a set of fine-grained permissions.
Root folderAn organization has an implicit top-level folder for each folder type (collection, watchlist, etc.).
TaskA request to run a single engine. A task is created with a payload containing input to the engine.
Temporal Data ObjectA wrapper that contains assets and associated time series and other metadata.
UserA single user account within an organization, representing a human user (not a service or application).
WatchlistAn enhanced stored search including a set of search conditions and filters, effective start and stop date, and other information. Hits against the search captured in a watchlist generate mentions.
Additional Technical Documentation Information
Properties
2/12/2024 10:21 PM
2/12/2024 10:22 PM
2/12/2024 10:22 PM
Documentation
Documentation
000004278
Translation Information
English

Powered by