Clarifai Guide
Clarifai Home
v7.0
v7.0
  • Welcome
  • Getting Started
    • Quick Start
    • Applications
      • Create an Application
      • Application Settings
      • Collaboration
    • Authentication
      • App-Specific API Keys
      • Personal Access Tokens
      • Scopes
      • Authorize
      • 2FA
    • Glossary
  • How-To
    • Portal
      • Auto Annotation
      • Custom Models
      • Text Classification
      • Visual Text Recognition
    • API
      • Auto Annotation
      • Batch Predict CSV on Custom Text Model
      • Custom KNN Face Classifier Workflow
      • Custom Models
      • Custom Text Model
      • Visual Text Recognition
  • API Guide
    • API overview
      • API Clients
      • Using Postman with Clarifai APIs
      • Status Codes
      • Pagination
      • Patching
    • Data Mode
      • Supported Formats
      • Create, Get, Update, Delete
      • Collectors
    • Concepts
      • Create, Get, Update
      • Languages
      • Search by Concept
      • Knowledge Graph
    • Scribe Label
      • Annotations
      • Training Data
      • Positive and Negative Annotations
      • Tasks
      • Task Annotations
    • Enlight Train
      • Clarifai Models
      • Model Types
      • Create, Get, Update, Delete
      • Deep Training
      • Evaluate
        • Interpreting Evaluations
        • Improving Your Model
    • Mesh Workflows
      • Base Workflows
      • Create, Get, Update, Delete
      • Input Nodes
      • Workflow Predict
    • Armada Predict
      • Images
      • Video
      • Prediction Parameters
      • Multilingual Classification
    • Spacetime Search
      • Search Overview
      • Combine or Negate
      • Filter
      • Rank
      • Index Images for Search
      • Legacy Search
        • Combine or Negate
        • Filter
        • Rank
        • Saved Searches
  • Portal Guide
    • Portal Overview
    • Data Mode
      • Supported Formats
      • Bulk Labeling
      • CSV and TSV
      • Collectors
    • Concepts
      • Create, Get, Update, Delete
      • Knowledge Graph
      • Languages
    • Scribe Label
      • Create a Task
      • Label Types
      • Labeling Tools
      • AI Assist
      • Workforce Management
      • Review
      • Training Data
      • Positive and Negative Annotations
    • Enlight Train
      • Training Basics
      • Clarifai Models
      • Model Types
      • Deep Training
      • Evaluate
        • Interpreting Evaluations
        • Improving Your Model
    • Mesh Workflows
      • Base Workflows
      • Setting Up a Mesh Workflow
      • Input Nodes
    • Armada Predict
    • Spacetime Search
      • Rank
      • Filter
      • Combine or Negate
      • Saved Searches
      • Visual Search
  • Data Labeling Services
    • Scribe LabelForce
  • Product Updates
    • Upcoming API Changes
    • Changelog
      • Release 7.0
      • Release 6.11
      • Release 6.10
      • Release 6.9
      • Release 6.8
      • Release 6.7
      • Release 6.6
      • Release 6.5
      • Release 6.4
      • Release 6.3
      • Release 6.2
      • Release 6.1
      • Release 6.0
      • Release 5.11
      • Release 5.10
Powered by GitBook
On this page

Was this helpful?

  1. Portal Guide
  2. Concepts

Knowledge Graph

Connect the knowledge of different models.

PreviousCreate, Get, Update, DeleteNextLanguages

Last updated 4 years ago

Was this helpful?

The Clarifai Knowledge Graph lets you map your custom concepts to a common set of concepts understood by all applications on the Clarifai platform. The knowledge graph makes it possible to link data across multiple custom and pre-built models in a meaningful way.

The Knowledge Graph uses Clarifai's concept mapping model to establish a hierarchical relationship between your concepts. Concept relationships are organized by three different predicates: hypernyms, hyponyms, and synonyms.

Hyponym represents an 'is a kind of' relation. The following relationship: 'honey' (subject), 'hyponym' (predicate), 'food' (object) is more easily be read as 'honey' 'is a kind of' 'food'.

Hypernym is the opposite of 'hyponym'. When you add one of the relationships the opposite will automatically appear for you in queries. The 'hypernym' can be read as 'is a parent of' so: 'food' (subject), 'hypernym' (predicate), 'honey' (object) can more easily be read as:'food' is a parent of 'honey'.

Synonym The 'synonym' relation defines two concepts that essential mean the same thing. This is more like a "is" relationship. So for example a 'synonym' relationship could be: "puppy" is "pup" The reverse is also true once the former is added so: "pup" is "puppy" will appear in queries as well.

You can create these mappings in your application with a few easy steps.

  1. Consider the following application that has four concepts: beverages, smoothie, breakfast, and french_toast. You can use the Knowledge Graph to create hierarchical relationships between the concepts.

  1. You can link concepts as hierarchical by going to the details of either of the concepts. In the shown application, french_toast falls under breakfast. You can link them by accessing the View Details section of either concept.

  1. Once, in the details dashboard, you can link breakfast as a hypernym to french_toast under the Input Relations menu.

  1. Once you list breakfast as a hypernym to french_toast, it will set french_toast as a hyponym to breakfast automatically.

  1. This process can be used to create similar relationships between beverages and smoothie. Beverages will be listed as a hypernym to smoothie.

  1. By doing this, smoothie will be listed as a hyponym to beverages automatically.