agrag.ingestion.extract.Extractor

Bases: ABC

Reads one chunk and returns the entities and relations it contains.

Note

Subclass this class to provide custom extraction. Graph awaits extract once for each chunk. GlinerExtractor is a local extractor. LLMExtractor marks the LLM role: Graph gives it only the weakest chunks, as llm_extractor. Graph drops any entity or relation whose label the schema does not declare, and logs the count.

Graph runs up to max_concurrency extract calls at the same time. An extractor that is not safe to call from several tasks at once sets max_concurrency to 1 or makes its calls take turns itself, for example with an asyncio.Lock.

Attributes:

• max_concurrency (int) – The most extract calls that run at once for one batch of chunks. Must be positive.

Functions:

• extract – Extract entities and relations from one chunk.

extract

extract(chunk:Chunk, schema:GraphSchema) -> ExtractionResult

Extract entities and relations from one chunk.

Parameters:

• chunk (Chunk) – The chunk to read. Only chunk.text and chunk.id are used.
• schema (GraphSchema) – The entity/relation types to extract. Every returned entity's label and relation's label must be declared in this schema.

Returns:

• ExtractionResult – The entities and relations this call found, in extraction order.

max_concurrency

max_concurrency: int = 8