How neo4j-graphacademy/llm-vectors-unstructured works
This is a GraphAcademy course repository demonstrating vector indexes and unstructured-data workflows with Neo4j, rather than a single deployed application. The main example loads AsciiDoc lessons, chunks them, creates embeddings and stores course structure in Neo4j; a query example embeds a question and searches a vector index. Topic extraction, vector-store setup, and Quora and poster embedding examples are additional branches. Several starter scripts are intentionally incomplete, and some branches are documented by the tree but not shown in source excerpts, so their relationships are omitted.
Main entry points: build_graph.py Load and chunk · build_graph_topics.py Topic graph writer · query_neo4j.py Vector query example · poster_embeddings.py Poster embeddings
Components 19
Course content
- Lesson corpus
llm-vectors-unstructured/data/asciidoc
Ingestion examples
- Load and chunk
llm-vectors-unstructured/solutions/build_graph.py - Course metadata
llm-vectors-unstructured/solutions/build_graph.py - Text embeddings
llm-vectors-unstructured/solutions/build_graph.py - Course graph writer
llm-vectors-unstructured/solutions/build_graph.py - Topic graph writer
llm-vectors-unstructured/solutions/build_graph_topics.py - Neo4j database
- Graph starter
llm-vectors-unstructured/build_graph.py - Embedding starter
llm-vectors-unstructured/create_embeddings.py
Vector search
- Vector query example
llm-vectors-unstructured/solutions/query_neo4j.py - Vector store setup
llm-vectors-unstructured/solutions/create_vector.py - Vector starter
llm-vectors-unstructured/create_vector.py - Query starter
llm-vectors-unstructured/query_neo4j.py
Additional examples
- Topic extraction demo
llm-vectors-unstructured/solutions/extract_topics.py - Quora filtering
llm-vectors-unstructured/solutions/quora_filter.py - Quora embeddings
llm-vectors-unstructured/solutions/quora_embeddings.py - Poster embeddings
llm-vectors-unstructured/solutions/poster_embeddings.py
Other components
- Course learner
- OpenAI embeddings
Connections 18
- Course learner → Lesson corpus: studies
README.adoc - Lesson corpus → Load and chunk: loads and chunks
llm-vectors-unstructured/solutions/build_graph.py - Load and chunk → Course metadata: passes chunks
llm-vectors-unstructured/solutions/build_graph.py - Course metadata → Text embeddings: requests embeddings
llm-vectors-unstructured/solutions/build_graph.py - Text embeddings → OpenAI embeddings: calls embeddings
llm-vectors-unstructured/solutions/build_graph.py - Course metadata → Course graph writer: supplies metadata
llm-vectors-unstructured/solutions/build_graph.py - Load and chunk → Course graph writer: dispatches writes
llm-vectors-unstructured/solutions/build_graph.py - Course graph writer → Neo4j database: writes graph
llm-vectors-unstructured/solutions/build_graph.py - Topic graph writer → Neo4j database: writes topics
llm-vectors-unstructured/solutions/build_graph_topics.py - Topic graph writer → OpenAI embeddings: requests embeddings
llm-vectors-unstructured/solutions/build_graph_topics.py - Vector query example → OpenAI embeddings: embeds question
llm-vectors-unstructured/solutions/query_neo4j.py - Vector query example → Neo4j database: queries index
llm-vectors-unstructured/solutions/query_neo4j.py - Vector store setup → Neo4j database: configures store
- Quora embeddings → OpenAI embeddings: requests embeddings
llm-vectors-unstructured/solutions/quora_embeddings.py - Poster embeddings → Neo4j database: reads movie posters
llm-vectors-unstructured/solutions/poster_embeddings.py - Poster embeddings → OpenAI embeddings: uses image model
- Graph starter → Neo4j database: planned graph write
- Query starter → Neo4j database: planned query