Appendix: Basic Terminology
Essential terms and their meaning in the context of AI Search.
| Term | Definition |
|---|---|
| AI Search | The Innovator Edge service that delivers semantic search across Innovator items and attached files. |
| Indexing Scope | The set of ItemTypes and file types an administrator has configured to be searchable. Defined in the AI Search Configurator. |
| Indexable Property | A property included in the semantic index for a given ItemType. The property selector in the configurator shows a consistent set of Innovator properties regardless of ItemType. Custom properties specific to individual ItemTypes, such as item_number or description, do not appear in the selector but can be added by specifying their internal name in the JSON schema and importing it. The indexing engine reads from the actual Innovator table, so any valid property name on that table can be indexed. |
| File Handling | The configuration that determines which file types attached to an ItemType’s items are extracted and included in search. Available only for ItemTypes that contain a property of type Item. The selectable file types are sourced live from your Innovator instance’s FileType table, not from a fixed product list. |
| File Properties | A section within each ItemType row in the AI Search Configurator, separate from File Handling Settings. File Properties controls which file-related fields are used for text extraction, vectorization, and search. Each row shows the field name, data type, and an enable/disable switch. Properties sourced from a relationship show a Rel badge with the relationship name. At least one file property must be enabled when file content indexing is required for an ItemType that has file handling configured. |
| Title Template | A configurable template controlling how a result card’s title is displayed, using {{ fieldName }} placeholders. Falls back to the item’s keyed_name if not set. |
| Subtitle Template | A configurable template controlling how a result card’s subtitle is displayed. Supports item property placeholders and {{score}}. |
| Relevance | A measure of how closely a result matches a search query, shown on each result card as a colored bar: green (High), blue (Medium), orange (Low), or gray (Lowest). |
| Staged filters | Filters that have been added, edited, or removed in the search UI but not yet sent to the search service. Staged filters appear as gray chips. They do not change search results until applied. |
| Applied filters | Filters that were included in the most recent search request. Applied filters appear as blue chips. Search results always reflect applied filters. |
| Indexing Status | The live-refreshing section of the AI Search Configurator showing the current state of each indexer in the active configuration, alongside timestamps, progress counts, and service health indicators. Refreshes every 30 seconds. Each indexer shows one of five states: Pending, Running, Completed, Partially Completed, or Failed. |
| Service Health Indicators | A row of status dots in the Indexing Status section showing the health of the six backend services AI Search depends on: sql-search, sql-data, innovator, embeddings, temporal, and cdc-configurator. A green dot indicates the service is healthy; a red dot indicates it is degraded or unhealthy. |
| Indexing Errors dialog | A dialog opened by selecting the N errors link on a Failed or Partially Completed indexer row. The dialog shows a summary of the run, a list of workflow errors with their timestamp, activity name, error category, message, and cause, expandable technical details, and a Copy error details action for support diagnostics. Error categories include Timeout, Activity Failure, and Application Failure. |
| Re-indexing | The process of rebuilding the index for an ItemType after its indexing scope configuration changes. Required whenever properties, file handling, file properties, or the set of indexed ItemTypes is changed on an instance that has already been indexed. Re-indexing behavior is determined by the backend automatically when Apply is submitted. |
| Change Data Capture (CDC) | The component that detects changes to indexed items and files in Innovator and triggers incremental re-indexing, keeping the index current after initial indexing completes. |
| Embedding | A vector representation of text or file content used by AI Search to match queries to relevant items by meaning rather than by exact keyword. |
| Semantic Search API | The API endpoint that returns ranked search results for a given query. Consumed by the AI Search web interface and available to external developers through Edge API. |
| Relationship ItemType | An ItemType that represents a relationship between two items, for example a BOM relationship. Relationship ItemTypes can be added directly to the indexing scope, or referenced as a source of file properties under a regular ItemType in the File Properties section. |
| Is Current | A search filter that limits results to the current version of an item. AI Search only indexes the current version of items and files; earlier versions are never included in the index. |
| Work Repository | The Aras DevOps / BDS source control repository from which a SaaS Innovator instance is built and deployed. For SaaS deployments, AI Search is enabled by adding files to the Work repository and running the deployment pipeline. |
| Chunk | A segment of text extracted from an indexed file during parsing. Chunks are the units that drive file-level search results. Each chunk carries a relevance score reflecting how closely it matches a given search query. Chunks are file chunks linked to an item through a property or relationship; item-level text is not chunked separately. |
| Chunk Retrieval | Access to the indexed file chunks that underlie search results. Available through the AI Search API in two forms: inline with each search response, controlled by the chunkCount parameter (0 to 20); or through a dedicated endpoint that returns all chunks for a specific item, ordered by document position and paginated by cursor. Both forms are access-controlled: only chunks for items the caller is authorized to see are returned. |
| Smart Configuration Changes Detection | The apply behavior introduced in version 1.1.0. When an administrator selects Apply, the AI Search service compares the saved and new configurations and re-indexes only the ItemTypes and content scopes affected by the difference. ItemTypes not affected remain fully searchable throughout the apply. No operation, including Apply and Reindex All, deletes file embeddings. |
| Apply and Reindex All | The secondary action on the Apply split button in the AI Search Configurator. Saves the current configuration, deletes all item embeddings for the active configuration, and triggers a full rebuild of item embeddings from source. File embeddings are not deleted or reprocessed by this operation. |