Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.xpertai.cn/llms.txt

Use this file to discover all available pages before exploring further.

The knowledge resource syncs knowledge base graphs with GraphRAG enabled in xpert-pro into the UOSE system. It is intended for read-only knowledge retrieval and evidence-based Q&A, not for re-extraction, entity writeback, or rebuild triggering.

Synced Objects

After sync, it usually contains:
  • knowledgebase: source knowledge base.
  • knowledge_entity: knowledge entity extracted or maintained in GraphRAG.
Core relationships include:
  • knowledgebase_has_entity
  • knowledge_entity_related
Entity attributes may include:
  • graph entity id.
  • entity type.
  • name.
  • origin.
  • visibility.
  • aliases.
  • description.
  • summary.
  • confidence.
  • mention count.
  • evidence samples.
  • metadata.

Prerequisites

Before registration, confirm that:
  • The xpert-pro knowledge base has GraphRAG enabled.
  • GraphRAG status is ready or rebuild_required.
  • The data-xpert API can access the xpert-pro API.
  • A token or API key that can call the knowledge base API is prepared.
If the status is disabled, indexing, or failed, sync fails and enters abnormal events.

Capabilities

Common capabilities include:
  • knowledgebaseId: required, source knowledge base ID.
  • includeHidden: whether to sync hidden graph items.
  • maxRelations: maximum number of synced relationships.
  • mentionSamplesPerItem: number of evidence samples kept per entity or relationship.
  • defaultRetrievalMode: default retrieval mode.
  • defaultTopK and maxTopK: retrieval count boundaries.
  • evidenceSamplesPerChunk: number of evidence samples per chunk.
These settings control context size and evidence density.

Actions

Knowledge base resources provide read-only graph actions, such as:
  • knowledge.graph.search_chunks
  • knowledge.graph.explore
  • knowledge.graph.neighborhood
  • knowledge.graph.entity_chunks
  • knowledge.graph.mentions
These actions are suitable for Agents to retrieve entities, inspect neighborhoods, and obtain evidence chunks and mentions.

Use Cases

Knowledge base resources are suitable for:
  • Providing traceable evidence in business Q&A.
  • Connecting entities and relationships from documents into the ontology workspace.
  • Letting Agents explore context along knowledge entity relationships.
  • Combining structured data analysis with unstructured document evidence.
When using them, control the sync volume of large GraphRAG graphs. Too many relationships and evidence samples increase context cost.