GML-2029 Remove the impractical embedding for Entity and other types

[chengbiao-jin]

PR Type

Enhancement, Bug fix, Documentation

---

Description

• Remove entity/relationship embeddings; simplify schema.

  • Rename Relationship to RelationshipType.
  • Drop ResolvedEntity, resolution workflow.
  • Keep vectors on DocumentChunk, Community.

• Safer vector upserts with schema discovery.

  • Cache vector attributes per vertex type.
  • Per-item vector_attribute support.

• Better multimodal defaults and providers.

  • Add Bedrock; sensible default vision models.

• Agent history contextualization and routing.

  • Merge KG results with conversation history.

---

Diagram Walkthrough

flowchart LR
  EMB["EmbeddingStore: vector attr cache + checks"]
  UPSERT["add_embeddings uses per-vertex attr"]
  SCHEMA["Schema: drop ResolvedEntity; use RelationshipType"]
  GSQL["GSQL queries updated accordingly"]
  AGENT["Agent: contextualize history, merge KG"]
  UI["UI/routers: clearer states and errors"]

  EMB -- "discover attrs via ls" --> UPSERT
  SCHEMA -- "affects" --> GSQL
  UPSERT -- "skip if attr missing" --> GSQL
  AGENT -- "history -> supportai -> merge" --> GSQL
  UI -- "status, selection, messages" --> AGENT

Loading

File Walkthrough

	Relevant files
Enhancement	17 files tigergraph_embedding_store.py Add vector-attribute cache and safe upserts                            +137/-17 config.py Validate configs; multimodal defaults; Bedrock support      +45/-18  graph_rag.py Remove entity resolution; streamline communities                  +7/-116  agent_graph.py Contextualize history and merge KG results                              +100/-9  supportai_ingest.py Stop embedding entities; use RelationshipType                        +9/-79    workers.py Drop embed channel; no entity embeddings                                  +4/-102  util.py Robust query installation with polling; concurrency            +44/-29  util.py Verify query install; poll until ready                                      +40/-16  image_data_extractor.py Cache multimodal client; simpler error texts                          +11/-4    base_llm.py Add prompt for question contextualization                                +15/-1    SupportAI_Schema.gsql Remove ResolvedEntity/Concept; add RelationshipType            +6/-16    SupportAI_Schema_Native_Vector.gsql Rename vector job; keep chunk/community vectors                    +2/-9      graphrag_louvain_init.gsql Run louvain directly on Entity vertices                                    +14/-14  GraphRAG_Hybrid_Search.gsql Adjust types; support RelationshipType context                      +2/-2      get_community_children.gsql Fetch children via Entity IN_COMMUNITY                                      +1/-1      Setup.tsx Track processing state; robust rebuild status                        +54/-29  Bot.tsx Validate selected graph; empty-state handling                        +25/-10
Bug fix	3 files eventual_consistency_checker.py Update upserts to RelationshipType; remove Concept links  +5/-25    inquiryai.py Clearer error messages for missing graphs                                +14/-6    text_extractors.py Sanitize alt text; lowercase image IDs; cleanup                    +12/-4
Documentation	1 files answer_question.py Update indices to chunks and communities                                  +1/-1
Additional files	54 files LICENSE +1/-1      html_chunker.py +1/-1      markdown_chunker.py +1/-1      recursive_chunker.py +1/-1      semantic_chunker.py +1/-1      connections.py +1/-1      LLMEntityRelationshipExtractor.py +1/-1      Build_Community_Concepts.gsql +0/-47    Build_Concept_Tree.gsql +0/-34    Build_Entity_Concepts.gsql +0/-28    Build_Relationship_Concepts.gsql +0/-25    getEntityRelationshipConceptCooccurrence.gsql +0/-140  ResolveRelationships.gsql +0/-26    StreamDocIds.gsql +0/-16    entities_have_resolution.gsql +0/-10    modularity.gsql +3/-3      Scan_For_Updates.gsql +1/-5      Selected_Set_Display.gsql +1/-1      SupportAI_IndexCreation.gsql +0/-3      SupportAI_Schema_Images.gsql +1/-1      Chunk_Sibling_Search.gsql +1/-1      Chunk_Sibling_Vector_Search.gsql +1/-1      Content_Similarity_Search.gsql +2/-2      Content_Similarity_Vector_Search.gsql +2/-2      Entity_Relationship_Retrieval.gsql +2/-2      GraphRAG_Community_Search.gsql +3/-3      GraphRAG_Community_Search_Display.gsql +3/-3      GraphRAG_Community_Vector_Search.gsql +4/-4      GraphRAG_Hybrid_Search_Display.gsql +2/-2      GraphRAG_Hybrid_Vector_Search.gsql +2/-2      Keyword_Search.gsql +1/-1      aws_bedrock_service.py +1/-1      google_genai_service.py +1/-1      openai_service.py +1/-1      schemas.py +1/-1      tool_io_schemas.py +1/-1      token_calculator.py +1/-1      community_summarizer.py +1/-1      main.py +2/-2      supportai_init.py +2/-2      workers.py +7/-14    agent.py +1/-1      agent_generation.py +1/-1      agent_router.py +1/-1      main.py +1/-1      supportai.py +1/-24    ui.py +7/-3      create_concepts.py +0/-103  supportai.py +2/-2      generate_cypher.py +1/-1      generate_function.py +1/-1      generate_gsql.py +1/-1      map_question_to_schema.py +1/-1      test_supportai.py +0/-6

---