AI System

Introduction

The AI System Schema is a JSON schema designed to define and standardise the representation of AI systems within the Trusted AI BOM (TAIBOM) ecosystem. This schema is applicable for systems used in both training AI models and inferencing, ensuring clarity and traceability of the underlying code and data.

Description

This schema captures essential metadata for an AI system, including:

• Name: The name of the AI system.
• Label: Indicates the system type, such as Training or Inferencing.
• Code: A reference to the code VC utilised within the system, with an optional cryptographic hash for verification.
• Data: A reference to the data VC, datapack, or weights used by the system, with an optional cryptographic hash for verification.

Use Case

The AI System Schema is primarily used within the TAIBOM framework to:

1. Define AI Systems: Provide a standardised representation of systems used for AI training or inferencing.
2. Track Dependencies: Ensure traceability by linking the system to its underlying code and data, along with optional integrity verification.
3. Enable Lifecycle Management: Support the management of AI systems across development, deployment, and maintenance phases.

By adopting this schema, organisations can ensure consistent documentation and robust lifecycle management of AI systems, contributing to transparency and accountability in AI ecosystems.

---

Schemas

yaml

$id: https://github.com/nqminds/Trusted-AI-BOM/blob/main/packages/schemas/src/taibom-schemas/50-ai-system.v1.0.0.schema.yaml
$schema: https://json-schema.org/draft/2019-09/schema
title: AI System
description: |
  This schema defines an AI system which may be used for training AI models or as an inferencing system.
type: object
properties:
  name:
    description: Name of the software system.
    type: string
  label:
    type: string
    description: Label of the system.
    enum:
      - Training
      - Inferencing
  code:
    type: object
    description: Reference to the code used in the system, with optional integrity verification.
    properties:
      id:
        type: string
        description: ID of the code VC.
      hash:
        type: string
        description: The cryptographic hash (e.g., SHA-256) of the code VC for integrity verification.
    required:
      - id
  data:
    type: object
    description: Reference to the datapack, data, or weights used on the system, with optional integrity verification.
    properties:
      id:
        type: string
        description: ID of the data VC.
      hash:
        type: string
        description: The cryptographic hash (e.g., SHA-256) of the data VC for integrity verification.
    required:
      - id
required:
  - name
  - label
  - code
  - data

json

{
  "$id": "https://github.com/nqminds/Trusted-AI-BOM/blob/main/packages/schemas/src/taibom-schemas/50-ai-system.v1.0.0.schema.yaml",
  "$schema": "https://json-schema.org/draft/2019-09/schema",
  "title": "AI System",
  "description": "This schema defines an AI system which may be used for training AI models or as an inferencing system.\n",
  "type": "object",
  "properties": {
    "name": {
      "description": "Name of the software system.",
      "type": "string"
    },
    "label": {
      "type": "string",
      "description": "Label of the system.",
      "enum": [
        "Training",
        "Inferencing"
      ]
    },
    "code": {
      "type": "object",
      "description": "Reference to the code used in the system, with optional integrity verification.",
      "properties": {
        "id": {
          "type": "string",
          "description": "ID of the code VC."
        },
        "hash": {
          "type": "string",
          "description": "The cryptographic hash (e.g., SHA-256) of the code VC for integrity verification."
        }
      },
      "required": [
        "id"
      ]
    },
    "data": {
      "type": "object",
      "description": "Reference to the datapack, data, or weights used on the system, with optional integrity verification.",
      "properties": {
        "id": {
          "type": "string",
          "description": "ID of the data VC."
        },
        "hash": {
          "type": "string",
          "description": "The cryptographic hash (e.g., SHA-256) of the data VC for integrity verification."
        }
      },
      "required": [
        "id"
      ]
    }
  },
  "required": [
    "name",
    "label",
    "code",
    "data"
  ]
}

markdown

AI System

This schema defines an AI system which may be used for training AI models or as an inferencing system.

The schema defines the following properties:

name (string, required)

Name of the software system.

label (string, enum, required)

Label of the system.

This element must be one of the following enum values:

• Training
• Inferencing

code (object, required)

Reference to the code used in the system, with optional integrity verification.

Properties of the code object:

id (string, required)

ID of the code VC.

hash (string)

The cryptographic hash (e.g., SHA-256) of the code VC for integrity verification.

data (object, required)

Reference to the datapack, data, or weights used on the system, with optional integrity verification.

Properties of the data object:

id (string, required)

ID of the data VC.

hash (string)

The cryptographic hash (e.g., SHA-256) of the data VC for integrity verification.

Examples

table

name	label	code	data
AI Model Training System	Training	[object Object]	[object Object]
AI Inferencing System	Inferencing	[object Object]	[object Object]

json

[
  {
    "name": "AI Model Training System",
    "label": "Training",
    "code": {
      "id": "urn:uuid:1cdf9f5c-b62c-4493-868a-code-abcdefg",
      "hash": "a1b2c3d4e5f6g7h8i9j0k1l2m3n4o5p6q7r8s9t0"
    },
    "data": {
      "id": "urn:uuid:1cdf9f5c-b62c-4493-868a-datapack-123",
      "hash": "z9y8x7w6v5u4t3s2r1q0p9o8n7m6l5k4j3i2h1g0"
    }
  },
  {
    "name": "AI Inferencing System",
    "label": "Inferencing",
    "code": {
      "id": "urn:uuid:1cdf9f5c-b62c-4493-868a-code-abcdefg",
      "hash": "f9e8d7c6b5a4d3c2b1a0e9f8d7c6b5a4d3c2b1a0"
    },
    "data": {
      "id": "urn:uuid:1cdf9f5c-b62c-4493-868a-datapack-456",
      "hash": "1a2b3c4d5e6f7g8h9i0j1k2l3m4n5o6p7q8r9s0t"
    }
  }
]

Edit this schema here