--- id: CON-PFM-QRY-001 name: Dataset feature: FEAT-PFM-QRY-01 product_area: platform module: queries revision: 1 status: draft product_intent: confirmed implementation_evidence: observed source_sha: services-platform: ad3f6a6582843008ed0f5a17b131878d15829af1 react-console: baf6dae382d6f41dce976c7f848667ccc37e076f definition: >- A dataset describes a kind of record available for questions, including its fields, relationships and access requirements as resolved for a workspace and reader. owned_by: MOD-PFM-QRY synonyms: - Data source - Record type (in the query designer) not_to_be_confused_with: - Source records maintained by their application - A saved question that selects values from a dataset - 'A physical table, whose relationship to a dataset need not be one-to-one' facets: - name: Fields meaning: >- Available values combine registered fields, selected collection details, calculated values and customer-defined information. used_for: 'Selecting what a question displays, compares and orders.' source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Relationships meaning: References and declared joins connect this kind of record to related kinds. used_for: Combining people with employers or summarizing related records. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Query permission meaning: >- The owning application supplies query authority, sometimes alongside record-read and application-access authority. used_for: Determining whether a reader may use the dataset. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Scopes and discovery rule meaning: >- Scopes choose an initial population, while discovery defaults identify records usually omitted from pickers. used_for: Supporting personal recent lists and ordinary record discovery. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Defaults meaning: 'Suggested fields, ordering and criteria provide a starting arrangement.' used_for: Giving a new question useful initial choices. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: System dataset meaning: HQ datasets describe platform operations across customers. used_for: Supporting Rhythm’s operational investigation. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Custom query table meaning: Rhythm staff publish a tailored read definition and assign its availability. used_for: Providing data shaped for selected customers. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Custom object dataset meaning: An active workspace-defined object supplies its own scoped dataset. used_for: Making customer-defined records available for questions. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented - name: Description meaning: A resolved description presents the current query vocabulary for a particular reader. used_for: Supplying choices and labels to query-building surfaces. source: 'services-platform:docs/queries/semantic-query-layer.md' provenance: documented relationships: - to: CON-PFM-QRY-002 statement: A question starts with one dataset and can include related datasets. provenance: documented source: 'services-platform:docs/queries/semantic-query-layer.md#4.2 The four concepts, kept distinct' - to: CON-PFM-QRY-005 statement: Quick Search projects its record types from their query datasets. provenance: documented source: 'services-platform:docs/quick-search.md#Quick Search: Operations' acted_on_by: - CAP-PFM-QRY-001 - CAP-PFM-QRY-002 - CAP-PFM-QRY-003 - CAP-PFM-QRY-004 - CAP-PFM-QRY-005 visible_to: - ACT-PFM-013 - ACT-PFM-014 - ACT-PFM-015 - ACT-PFM-016 evidence: - 'Pinned reference: services-platform:docs/queries/semantic-query-layer.md' - 'Pinned reference: services-platform:docs/quick-search.md' supersedes: [] origin: 'services-platform:docs/queries/semantic-query-layer.md#2026-10-08 — Final acceptance' intent_sources: - 'services-platform:docs/queries/semantic-query-layer.md#2026-10-08 — Final acceptance' --- # Dataset ## What It Is A dataset is the shared description of a kind of record that people can investigate. Contacts, organizations and accounting lines are examples. Staff encounter it as a choice of subject for a question, together with the fields and relationships that subject offers. Its identity follows the application and declared record kind, rather than the label a customer happens to display. The owning application supplies the declaration. Platform brings those declarations together and resolves them with the workspace’s metadata and the reader’s access. This keeps the query vocabulary separate from any particular screen. A customer-defined label or additional field can become part of the description without changing the identity of the subject being queried. A dataset therefore gives Product a useful boundary: which records can be investigated, what can be asked about them, and which authority permits that investigation. It does not transfer ownership of the records to Platform. ## What It Is Not A dataset differs from the stored records it describes. It also differs from a physical table: a record description can combine several sources or select a detail such as a preferred address. A saved question selects and filters that vocabulary. A report presents the result of saved questions. These distinctions matter because changing a presentation does not change the available data model or its owner. ## How It Is Used Staff browse datasets before building a question. The catalog includes unauthorized entries, making missing access visible without making those records queryable. An available dataset supplies the choices used by the designer and related list tools. Applications register their declarations during deployment and receive errors when a declaration cannot be accepted. Rhythm staff use a separate system catalog for platform operations. They also maintain custom definitions for selected customers. Customers can activate custom objects whose datasets then become available within their workspace. All three forms provide a query subject, but their ownership and availability differ. ## Its Facets **Fields.** Values include ordinary record fields, custom information, calculations and selected collection details. Their types and supported operations determine how a question can use them. **Relationships.** References and declared joins connect records without requiring the person to understand their storage. The reader’s access affects which related vocabulary is available. **Query permission.** Dataset authority comes from its owner. Record-read and application access can add requirements beyond the query permission. **Scopes and discovery rule.** A scope supplies an initial population, such as recently viewed records. Discovery defaults guide initial selection rather than silently changing submitted questions. **Defaults.** Suggested columns, criteria and ordering make the first arrangement useful while remaining distinct from the actual request. **System dataset.** HQ subjects describe platform records across customers and remain separate from customer record subjects. **Custom query table.** A published tailored definition appears as a dataset where its assignments allow it. Its lifecycle and publication health affect continued use. **Custom object dataset.** Workspace-defined records use their own scoped access and metadata. **Description.** The resolved description supplies current labels and choices to clients, including changes that invalidate a cached description. ## How It Relates Queries name datasets; reports and scheduled files depend on those queries. Search uses dataset projections to build a separate index. Export coverage records declared extraction subjects separately. CRM, Accounting and other applications remain responsible for the records being described. ## What It Could Become The product should decide how future field and record restrictions fit this vocabulary. Tailored datasets also need clear authority for later administration features. Any expansion should preserve the distinction between describing records, permitting access and presenting a result.