> ## Documentation Index > Fetch the complete documentation index at: https://docs.hebbianrobotics.com/llms.txt > Use this file to discover all available pages before exploring further. # Get dataset content diversity > Returns canonical environment, visible-object, and action-verb concepts, together with object interaction and action-verb/object distributions across analyzed clips. Descriptive categories are included as optional ways to group the concepts. This endpoint characterizes content composition; it does not directly detect duplicate video or determine task completion. Use `axis` to request one independent concept type. For field examples and comparison guidance, see [Understand quality metrics](https://docs.hebbianrobotics.com/quality-metrics). ## OpenAPI ````yaml https://api.hebbianrobotics.com/public-openapi.json get /api/v1beta/quality/content-diversity openapi: 3.1.0 info: description: >- Measure video characteristics, hand activity, and content diversity in robotics video datasets. Start with `GET /api/v1beta/access`, then list available datasets with `GET /api/v1beta/quality/datasets`. license: identifier: Apache-2.0 name: Apache-2.0 title: Robotics Data Quality API version: v1beta servers: - description: Production API url: https://api.hebbianrobotics.com security: - {} - bearer_token: [] tags: - description: >- Video-characteristics, hand-activity, and content-diversity measurements for robotics video data. name: quality - description: Scope and capabilities granted by a bearer token. name: access paths: /api/v1beta/quality/content-diversity: get: tags: - quality summary: Get dataset content diversity description: >- Returns canonical environment, visible-object, and action-verb concepts, together with object interaction and action-verb/object distributions across analyzed clips. Descriptive categories are included as optional ways to group the concepts. This endpoint characterizes content composition; it does not directly detect duplicate video or determine task completion. Use `axis` to request one independent concept type. For field examples and comparison guidance, see [Understand quality metrics](https://docs.hebbianrobotics.com/quality-metrics). operationId: getContentDiversity parameters: - description: Dataset ID from `GET /api/v1beta/quality/datasets`. in: query name: dataset required: true schema: type: string - description: >- Opaque `content_diversity` token from an earlier response. Omit to use the latest measurement. in: query name: content_diversity required: false schema: type: string - description: |- Return only one diversity axis: `environments`, `objects`, or `action_verbs`. Omit it to receive all three independent distributions. in: query name: axis required: false schema: $ref: '#/components/schemas/QualityDiversityAxisQuery' responses: '200': content: application/json: schema: $ref: '#/components/schemas/ContentDiversityResponse' description: Published concepts and clip distributions by diversity axis headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '400': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: The dataset ID or axis is invalid headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '401': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: A valid bearer token is required headers: WWW-Authenticate: description: Bearer authentication challenge schema: type: string x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '403': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: The bearer token cannot read quality measurements headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '404': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: >- The dataset, content-diversity measurement, or a source measurement is unavailable headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '409': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: >- The requested content-diversity measurement is stale; omit the content_diversity token and retry headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '500': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: The server could not read content-diversity measurements headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '503': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: The server is temporarily unavailable or overloaded headers: Retry-After: description: Seconds to wait before retrying an overloaded request schema: type: string x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string '504': content: application/json: schema: $ref: '#/components/schemas/ApiErrorBody' description: The request exceeded the server processing deadline headers: x-request-id: description: >- Request correlation identifier supplied by the caller or generated by the API schema: type: string security: - bearer_token: [] components: schemas: QualityDiversityAxisQuery: enum: - environments - objects - action_verbs type: string ContentDiversityResponse: description: >- Dataset content diversity and measurement completeness. Use this response to inspect which concepts are present and how frequently they occur across analyzed clips. It does not directly detect duplicate video or determine whether a task was completed correctly. properties: analyzed_clip_count: description: >- Number of distinct clips that produced usable content observations. This is the denominator for every `clip_percent` in `axes` and `object_actions`. minimum: 0 type: integer axes: description: >- Independent distributions for environment, object, and action-verb concepts. An `axis` query parameter limits this array to one requested type. items: $ref: '#/components/schemas/QualityDiversityAxisDto' type: array content_diversity: description: >- Opaque content-diversity token. Matching tokens identify results from the same content-diversity measurement. type: string coverage_percent: description: >- Percentage of the planned analysis that successfully produced values used in this response, in `[0, 100]`. This is measurement completeness, not the percentage of footage that passed quality control. `null` when the planned result count cannot be determined reliably. Non-null values are rounded to two decimal places. format: double maximum: 100 minimum: 0 type: - number - 'null' dataset: description: Dataset ID used in the request. type: string object_actions: $ref: '#/components/schemas/QualityObjectActionsDto' description: >- Observed action-verb and object combinations. Independent object and action-verb distributions are derived from the same activity records. required: - dataset - content_diversity - coverage_percent - analyzed_clip_count - axes - object_actions type: object ApiErrorBody: description: Error payload returned for handled API failures. properties: code: description: Machine-readable error code, e.g. `dataset_not_found`. type: string message: description: Human-readable message. type: string required: - code - message type: object QualityDiversityAxisDto: description: |- Concept count and clip distribution for environments, objects, or action verbs. properties: axis: description: >- Type of concept summarized by this object: `environments`, `objects`, or `action_verbs`. Each axis is measured independently. type: string categories: description: >- Every possible descriptive category for this axis, including categories with no matching concepts. Use each returned `label` when presenting it. items: $ref: '#/components/schemas/QualityConceptCategoryDto' type: array concept_count: description: |- Number of canonical concepts published for this axis. Each concept counts once no matter how many clips contain it. minimum: 0 type: integer concepts: description: >- Every published concept, most frequent first, with its clip distribution. items: $ref: '#/components/schemas/QualityConceptDto' type: array required: - axis - concept_count - categories - concepts type: object QualityObjectActionsDto: description: Distribution of observed action-verb and object combinations. properties: combination_count: description: Number of canonical action-verb and object combinations. minimum: 0 type: integer combinations: description: Every published combination, most frequent first. items: $ref: '#/components/schemas/QualityObjectActionDto' type: array required: - combination_count - combinations type: object QualityConceptCategoryDto: description: |- Summary of one descriptive category within a diversity axis. Environment categories group setting types. Object and action-verb categories group concepts by handling or dexterity characteristics. A category is a descriptive bucket, not a quality score. properties: category: description: Stable machine-readable category name. type: string concept_count: description: Number of published concepts assigned to this category. minimum: 0 type: integer example_concepts: description: Frequent concepts that make the category concrete for a reviewer. items: type: string type: array label: description: |- Display category name. Prefer this value in reports and user interfaces. type: string required: - category - label - concept_count - example_concepts type: object QualityConceptDto: description: >- One canonical environment, object, or action verb observed in the footage. properties: category: description: |- Machine-readable category assigned to the concept. `null` when no category was assigned. A category is not a quality score. type: - string - 'null' clip_count: description: >- Number of distinct analyzed clips containing the concept. A concept is counted at most once per clip. minimum: 0 type: integer clip_percent: description: >- Percentage of analyzed clips containing the concept, in `[0, 100]`. Concepts can co-occur, so percentages within an axis do not need to sum to 100. Values are rounded to two decimal places. format: double maximum: 100 minimum: 0 type: number concept: description: Canonical concept published by the API, such as `cup`. type: string interaction: oneOf: - type: 'null' - $ref: '#/components/schemas/QualityObjectInteractionDto' description: >- Clip distribution where this object was visibly involved in an action. Present only for concepts in the `objects` axis. required: - concept - clip_count - clip_percent - category type: object QualityObjectActionDto: description: >- One canonical action-verb and object combination observed in the footage. properties: action_verb: description: Canonical action verb, such as `pick up`. type: string clip_count: description: |- Number of distinct analyzed clips containing this combination. A combination is counted at most once per clip. minimum: 0 type: integer clip_percent: description: >- Percentage of analyzed clips containing this combination, in `[0, 100]`. Combinations can co-occur, so percentages do not need to sum to 100. Rounded to two decimal places. format: double maximum: 100 minimum: 0 type: number object: description: Canonical object or material involved in the action, such as `cup`. type: string required: - action_verb - object - clip_count - clip_percent type: object QualityObjectInteractionDto: description: How often a visible object was also involved in an observed action. properties: clip_count: description: >- Number of distinct analyzed clips where the object was involved in at least one action. An object is counted at most once per clip. minimum: 0 type: integer clip_percent: description: >- Percentage of analyzed clips where the object was involved in at least one action, in `[0, 100]`. Rounded to two decimal places. format: double maximum: 100 minimum: 0 type: number required: - clip_count - clip_percent type: object securitySchemes: bearer_token: description: Bearer access token issued through OAuth 2.0 client credentials. scheme: bearer type: http ````