Query-focused Summarization

Query-focused summarization (QFS) is a text analysis process that generates a concise summary of documents specifically tailored to address a user’s query or topic of interest. Unlike generic summarization that covers the overall main points, QFS filters, prioritizes, and compresses information relevant to the user’s specific information needs.

Prerequisites

homogeneous graph
each node having the text property

How to do QFS?

When working with a knowledge graph, you can create summaries based on its underlying communities. Keep in mind: community detection works best on homogeneous graphs, that is, graphs where nodes of the same type (with the same label) are connected by edges. This structure ensures that the detected communities are meaningful. Once your graph is structured this way, you can detect and assign communities using queries such as the following (multiple community detection algorithms are possible):

CALL community_detection.get() YIELD node, community_id
SET node.community_id = community_id;

After assigning community IDs, you can generate summaries for each community and create corresponding Community nodes. These nodes are then connected to their community members using the Cypher query below:

WITH "You are summarizing a community (cluster)." +
     " Write a concise summary in 2–4 paragraphs that:" +
     " - Captures the main themes and recurring topics."
     " - Describes the kind of work or concerns this cluster represents." +
     " - Does not list individual entities; synthesize into a coherent narrative."
     AS SUMMARY_TEMPLATE
MATCH (n) WHERE n.community_id IS NOT NULL
WITH n.community_id AS c_id, count(n) AS c_count, collect(n) AS c_members,
     llm.complete(reduce(s=SUMMARY_TEMPLATE, m IN collect(n) | s + m.text "; ")) AS c_summary
MERGE (community:Community {id: c_id, nodes_count: c_count})
SET community.summary = c_summary
WITH community, c_members
UNWIND c_members AS c_member
MERGE (c_member)-[:BELONGS_TO]->(community)
RETURN community.id AS community_id, community.summary AS community_summary;

Once all the preprocessing is done, a typical query to perform query-focused summarization follows:

WITH "... the question ..." AS USER_QUESTION
// 1. Identify relevant communities (thresholding)
MATCH (c:Community) WHERE c.nodes_count > 5  // only use significant communities
// 2. Map: Generate a partial answer for EACH community
WITH USER_QUESTION, c,
     llm.complete("How does the following community summary: " + c.summary +
                  " relate to: " + USER_QUESTION) AS partial_answer
  WHERE partial_answer IS NOT NULL AND partial_answer <> ""
WITH USER_QUESTION, collect(partial_answer) AS map_results
// 3. Reduce: Synthesize the final consolidated answer
WITH "The following are partial answers from different thematic clusters of " +
     " a dataset. Synthesize them into a single, cohesive response to the " +
     "original query: " + USER_QUESTION +
      "\n\nPartial Answers:\n" +
      reduce(s = "", res IN map_results | s + "- " + res + "\n") AS reduce_prompt
RETURN llm.complete(reduce_prompt) AS final_answer;

Local Graph Search Agentic GraphRAG