Skip to main content
GraphRAG uses several enumeration types to constrain configuration values to valid options.

​
ReportingType

The reporting configuration type for the pipeline. Module: graphrag.config.enums Base class: str, Enum

​
Values

​
file
str
default:"file"
The file reporting configuration type. Writes reporting output to local files.
​
blob
str
default:"blob"
The blob reporting configuration type. Writes reporting output to Azure Blob Storage.

​
Example

from graphrag.config.enums import ReportingType

# Use in configuration
reporting_type = ReportingType.file
print(reporting_type)  # "file"

​
AsyncType

Enum for the type of async to use for concurrent operations. Module: graphrag.config.enums Base class: str, Enum

​
Values

​
AsyncIO
str
default:"asyncio"
Use Python’s asyncio for asynchronous operations. Provides true async/await concurrency.
​
Threaded
str
default:"threaded"
Use thread-based concurrency for asynchronous operations. Default mode for most operations.

​
Example

from graphrag.config.enums import AsyncType

# Use in configuration
async_mode = AsyncType.Threaded

​
SearchMethod

The type of search to run. Module: graphrag.config.enums Base class: Enum

​
Values

​
LOCAL
str
default:"local"
Local search method. Searches within a local context using entities and relationships.
​
GLOBAL
str
default:"global"
Global search method. Performs map-reduce search across community reports.
​
DRIFT
str
default:"drift"
DRIFT search method. Dynamic reasoning and inference with followup trajectories.
​
BASIC
str
default:"basic"
Basic search method. Simple vector similarity search.

​
Example

from graphrag.config.enums import SearchMethod

# Use string representation
search_type = SearchMethod.LOCAL
print(str(search_type))  # "local"

​
IndexingMethod

Enum for the type of indexing to perform. Module: graphrag.config.enums Base class: str, Enum

​
Values

​
Standard
str
default:"standard"
Traditional GraphRAG indexing, with all graph construction and summarization performed by a language model. Provides highest quality but slower performance.
​
Fast
str
default:"fast"
Fast indexing, using NLP for graph construction and language model for summarization. Significantly faster than standard indexing.
​
StandardUpdate
str
default:"standard-update"
Incremental update with standard indexing. Updates an existing index with new documents using LLM-based extraction.
​
FastUpdate
str
default:"fast-update"
Incremental update with fast indexing. Updates an existing index with new documents using NLP-based extraction.

​
Example

from graphrag.config.enums import IndexingMethod

# Use in indexing configuration
method = IndexingMethod.Fast

​
NounPhraseExtractorType

Enum for the noun phrase extractor options used in NLP-based graph extraction. Module: graphrag.config.enums Base class: str, Enum

​
Values

​
RegexEnglish
str
default:"regex_english"
Standard extractor using regex. Fastest option, but limited to English language text.
​
Syntactic
str
default:"syntactic_parser"
Noun phrase extractor based on dependency parsing and named entity recognition (NER) using SpaCy. More accurate than regex but slower.
​
CFG
str
default:"cfg"
Noun phrase extractor combining context-free grammar (CFG) based noun-chunk extraction and NER. Balanced approach between speed and accuracy.

​
Example

from graphrag.config.enums import NounPhraseExtractorType

# Use in NLP extraction configuration
extractor = NounPhraseExtractorType.RegexEnglish

​
ModularityMetric

Enum for the modularity metric to use in graph clustering. Module: graphrag.config.enums Base class: str, Enum

​
Values

​
Graph
str
default:"graph"
Graph modularity metric. Standard modularity calculation for community detection.
​
LCC
str
default:"lcc"
Largest connected component modularity metric.
​
WeightedComponents
str
default:"weighted_components"
Weighted components modularity metric. Takes edge weights into account when calculating modularity.

​
Example

from graphrag.config.enums import ModularityMetric

# Use in clustering configuration
metric = ModularityMetric.Graph

​
Usage in configuration

Enums are used throughout the GraphRAG configuration to provide type safety and validation: 
# settings.yaml example
reporting:
  type: file  # Uses ReportingType.file
  base_dir: logs

async_mode: threaded  # Uses AsyncType.Threaded

extract_graph_nlp:
  text_analyzer:
    extractor_type: regex_english  # Uses NounPhraseExtractorType.RegexEnglish
When loading configuration, string values are automatically converted to the appropriate enum types.

Build docs developers (and LLMs) love

Get started for freeTalk to us