Community summary

As explained in Essential GraphRAG

A community-level summary is a concise natural-language description of the entities, relationships, themes, and information associated with a graph community.

A community summary is an LLM-generated, high-level account of a densely connected group of graph entities. It consolidates entity and relationship information into a cohesive intermediate representation for retrieval and response generation. section 7.0 section 7.4

Community summaries support GraphRAG retrieval in several ways:

• Global search retrieves reports from a selected community-hierarchy level, processes report chunks in a map step, filters and aggregates important points, and gives the resulting context to a final LLM. section 7.3.1
• Local search can add summaries from communities connected to vector-matched entities, combining them with chunks, relationships, and entity descriptions. section 7.3.2
• The chosen community level trades detail against efficiency: lower levels improve thoroughness but require more LLM calls and processing time, while higher levels risk losing granularity. section 7.3.1

Global question answering

A global retriever queries stored __Community__ nodes, selects summaries above a rating threshold, asks an LLM to extract rated key points from each summary, and synthesizes the results for a broad query such as “What is this story about?”

Figure 7.4 Global search

Use the global-search diagram to trace community reports through map processing, point aggregation, and final answer synthesis. section 7.3.1

How Essential GraphRAG covers it

How deep it goes

Explains generation, hierarchy selection, map-reduce use, and graph integration, but does not detail the summarization prompt or algorithm internals.

Examples

Demonstrates stored community reports, rated key points, map-reduce aggregation, a global retriever, and local-search context for Ulysses.

What it assumes

Assumes extracted entities, relationships, a graph index, community detection, and LLM-based summarization; introduces their coordinated retrieval role.

What you can do after

Apply community summaries in global or local GraphRAG retrieval and tune hierarchy level or rating thresholds for efficiency and detail.

How it connects

Connects community detection, knowledge graphs, entity summaries, relationship summaries, global search, local search, and vector similarity search.

liveBookmanning.com

Read Essential GraphRAG on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  Domain-Specific Small Language ModelsA high-level summary of a group of related graph entities used as an intermediate unit for search and response generation.

Compare how 2 books explain community summary →

Related concepts

• global searchused by · Global search is a GraphRAG retrieval technique that answers broad queries by aggregating and synthesizing information across community summaries or other corpus-level representations.
• GraphRAGproduced by · GraphRAG is a retrieval-augmented generation approach that uses graph-structured knowledge and relationships to retrieve context for language-model responses.
• MapReduceused by · MapReduce is a computational pattern that applies map operations to input elements or partitions and combines the intermediate results through reduction or aggregation.
• language modelproduced by · A language model is a computational model that represents patterns in linguistic tokens to estimate sequence probabilities and predict or generate language from context.
• knowledge graphhas part · A knowledge graph is a graph-structured representation of entities and their semantically meaningful relationships, often including associated attributes or facts.

This explainer was generated by Manning's knowledge graph from the text of Essential GraphRAG. The section links point to the parts of the book it draws on.