Building Knowledge Graphs from Unstructured Text with Claude

You have a pile of unstructured documents and need to answer questions that span them — "who works with people who worked on project X", "which vendors are connected to this incident". No single document contains the answer. RAG retrieval won't chain the facts for you. You need a knowledge graph: entities as nodes, typed relations as edges, so that multi-hop reasoning becomes graph traversal.

Building one used to mean training a named-entity recognizer on your domain, training a relation classifier, writing entity-resolution heuristics, and maintaining all three as your data shifted. With Claude, each of those stages becomes a prompt.

What You'll Learn

By the end of this guide you will be able to:

• Use structured outputs to extract typed entities and subject–predicate–object triples from arbitrary text with no training data
• Apply Claude-driven entity resolution to collapse surface-form variants into canonical nodes, replacing brittle string-similarity heuristics
• Assemble and query an in-memory graph, and run multi-hop questions by serializing subgraphs back to Claude
• Measure extraction quality with precision/recall against a gold set and reason about the cost/quality tradeoff between Haiku and Sonnet

Everything runs in memory with no database. The techniques transfer directly to Neo4j, Neptune, or a Postgres adjacency table when you need to scale.

Prerequisites

• Python 3.11+
• Anthropic API key (get one here)
• Basic familiarity with graphs (nodes, edges, traversal)

Setup

We use two models. Haiku handles the high-volume, schema-constrained extraction work where speed and cost matter more than nuance. Sonnet handles entity resolution and summarization, where the model needs to weigh conflicting evidence across documents.

import anthropic
from pydantic import BaseModel, Field
from typing import List, Optional
client = anthropic.Anthropic()
Define your extraction schema
class Entity(BaseModel):
    name: str = Field(description="The canonical name of the entity")
    type: str = Field(description="Entity type: PERSON, ORG, LOC, EVENT, etc.")
    description: str = Field(description="One-line description for disambiguation")
class Relation(BaseModel):
    subject: str = Field(description="Name of the subject entity")
    predicate: str = Field(description="Relation type (e.g., 'works_at', 'launched')")
    object: str = Field(description="Name of the object entity")
class Extraction(BaseModel):
    entities: List[Entity]
    relations: List[Relation]

Building a Corpus

We need a handful of documents that talk about overlapping entities, so that entity resolution has real work to do. The Apollo program is a good test bed: six short Wikipedia summaries that all mention NASA, the Moon, several astronauts, and a launch vehicle — but each article names them slightly differently.

We fetch summaries from the Wikipedia REST API rather than full articles to keep token costs low. For a production pipeline you would chunk full documents; the extraction logic is identical.

import requests
def fetch_wikipedia_summary(title):
    url = f"https://en.wikipedia.org/api/rest_v1/page/summary/{title}"
    response = requests.get(url)
    response.raise_for_status()
    return response.json()["extract"]
documents = {
    "Apollo 11": fetch_wikipedia_summary("Apollo 11"),
    "Neil Armstrong": fetch_wikipedia_summary("Neil Armstrong"),
    "Buzz Aldrin": fetch_wikipedia_summary("Buzz Aldrin"),
    "Saturn V": fetch_wikipedia_summary("Saturn V"),
    "NASA": fetch_wikipedia_summary("NASA"),
    "Moon landing": fetch_wikipedia_summary("Moon landing")
}

Entity and Relation Extraction

Classical NER tags spans of text with labels (PERSON, ORG, LOC). Classical relation extraction then classifies pairs of spans into relation types. Both traditionally require labeled training data per domain.

We collapse both stages into a single Claude call per document. The key is structured outputs: we define the output shape as a Pydantic model and pass it to client.messages.parse(). Claude's response is guaranteed to validate against that schema and comes back as a typed Python object — no regex parsing, no JSON decode errors, no defensive isinstance checks.

def extract_from_document(title, text):
    response = client.messages.parse(
        model="claude-3-haiku-20240307",
        max_tokens=2000,
        system="You are a knowledge graph extraction specialist. Extract all entities and their relations from the text.",
        messages=[
            {
                "role": "user",
                "content": f"Extract entities and relations from this document titled '{title}':\n\n{text}"
            }
        ],
        response_model=Extraction
    )
    return response
Extract from all documents
all_extractions = {}
for title, text in documents.items():
    all_extractions[title] = extract_from_document(title, text)

Let's look at what was extracted. Notice how the same real-world entity appears under different surface forms across documents — this is the entity resolution problem we solve next.

Entity Resolution

The raw extraction gives us overlapping mentions: "NASA" and "National Aeronautics and Space Administration", "Neil Armstrong" and "Armstrong", possibly "the Moon" and "Moon". If we build a graph directly from this, we get a fractured mess where the same concept is split across disconnected nodes.

Traditional approaches use string similarity (edit distance, Jaccard on tokens) plus blocking rules. That works for typos but fails on "Edwin Aldrin" vs "Buzz Aldrin" — two names with zero character overlap that refer to the same person.

We instead ask Claude to cluster entities of each type, using the one-line descriptions from extraction as disambiguation context. The descriptions matter: "Armstrong — first person to walk on the Moon" and "Armstrong — jazz trumpeter" have the same name but should not merge.

def resolve_entities(extractions):
    # Collect all unique entities with their descriptions
    entity_registry = {}
    for doc_title, extraction in extractions.items():
        for entity in extraction.entities:
            key = (entity.name, entity.type)
            if key not in entity_registry:
                entity_registry[key] = entity.description
    
    # Ask Claude to cluster by type
    resolved = {}
    for entity_type in set(e.type for e in entity_registry):
        type_entities = {k: v for k, v in entity_registry.items() if k[1] == entity_type}
        
        response = client.messages.parse(
            model="claude-3-sonnet-20240229",
            max_tokens=2000,
            system="You are an entity resolution specialist. Group entity names that refer to the same real-world entity.",
            messages=[
                {
                    "role": "user",
                    "content": f"""Group these {entity_type} entities into clusters where each cluster represents one real-world entity.
                    Return a list of clusters, each with a canonical name and the list of aliases.
                    
                    Entities and descriptions:
                    {chr(10).join(f'- {name}: {desc}' for (name, _), desc in type_entities.items())}
                    """
                }
            ],
            response_model=List[Cluster]
        )
        
        for cluster in response:
            for alias in cluster.aliases:
                resolved[(alias, entity_type)] = cluster.canonical_name
    
    return resolved

Two failure modes to watch for. First, any raw name Claude leaves out of every cluster silently disappears from the graph — a production resolver should fall back to a single-element cluster for unmatched names so nothing is lost. Second, the resolver can over-merge: a specific mission like "Gemini 12" may get folded into the broader "Project Gemini" because the descriptions overlap. The first loses nodes, the second loses precision. Both are worth spot-checking in the output below.

Assembling the Graph

With a clean alias map, we rewrite every relation endpoint to its canonical form and load the result into NetworkX. We use a MultiDiGraph because two entities can be connected by several distinct predicates ("launched from" and "operated by"), and direction matters ("Armstrong commanded Apollo 11" is not the same edge as "Apollo 11 commanded Armstrong").

Each node carries its type, the source document it came from, and the original surface form for auditability.

import networkx as nx
def build_graph(extractions, alias_map):
    G = nx.MultiDiGraph()
    
    for doc_title, extraction in extractions.items():
        for entity in extraction.entities:
            canonical = alias_map.get((entity.name, entity.type), entity.name)
            G.add_node(
                canonical,
                type=entity.type,
                source=doc_title,
                original_name=entity.name
            )
        
        for relation in extraction.relations:
            subj_canon = alias_map.get((relation.subject, ""), relation.subject)
            obj_canon = alias_map.get((relation.object, ""), relation.object)
            G.add_edge(
                subj_canon,
                obj_canon,
                predicate=relation.predicate,
                source=doc_title
            )
    
    return G
graph = build_graph(all_extractions, alias_map)
print(f"Graph has {graph.number_of_nodes()} nodes and {graph.number_of_edges()} edges")

Querying the Graph with Multi-Hop Reasoning

Now the real power: answering questions that require traversing multiple edges. We serialize a relevant subgraph back to Claude and let it reason over the structure.

def query_graph(question, graph, max_hops=2):
    # Serialize the graph as a list of triples
    triples = []
    for u, v, data in graph.edges(data=True):
        triples.append(f"({u}) --[{data['predicate']}]--> ({v})")
    
    graph_context = "\n".join(triples)
    
    response = client.messages.create(
        model="claude-3-sonnet-20240229",
        max_tokens=1000,
        system="You are a knowledge graph query engine. Answer questions by traversing the graph.",
        messages=[
            {
                "role": "user",
                "content": f"""Given this knowledge graph:
{graph_context}
Answer the following question by reasoning over the graph:
{question}
Explain your reasoning step by step."""
            }
        ]
    )
    return response.content[0].text
Example multi-hop query
answer = query_graph(
    "Which astronauts were involved in missions that used the Saturn V rocket?",
    graph
)
print(answer)

This approach lets you answer questions like:

• "Who worked with people who worked on Project Apollo?"
• "Which vendors are connected to this incident?"
• "What organizations were involved in the Moon landing?"

Measuring Quality

To trust your graph, you need to measure precision and recall against a gold standard. Create a small set of known entities and relations for a subset of your documents, then compare:

def evaluate_extraction(gold_entities, gold_relations, extracted_entities, extracted_relations):
    # Precision: what fraction of extracted items are correct?
    # Recall: what fraction of gold items were extracted?
    
    true_positives_entities = len(gold_entities & extracted_entities)
    precision_entities = true_positives_entities / len(extracted_entities) if extracted_entities else 0
    recall_entities = true_positives_entities / len(gold_entities) if gold_entities else 0
    
    # Similar for relations
    
    return {
        "entity_precision": precision_entities,
        "entity_recall": recall_entities,
        "f1_entities": 2  (precision_entities  recall_entities) / (precision_entities + recall_entities) if (precision_entities + recall_entities) > 0 else 0
    }

Cost/Quality Tradeoffs

Haiku is ideal for high-volume extraction where speed and cost matter. Sonnet excels at entity resolution and complex reasoning where nuance is critical. In production, use Haiku for initial extraction and Sonnet for resolution and querying.

Key Takeaways

• Structured outputs eliminate parsing headaches: Define your schema with Pydantic and let Claude return validated objects directly — no regex, no JSON parsing errors.
• Claude handles entity resolution better than string similarity: It understands that "Buzz Aldrin" and "Edwin Aldrin" are the same person, even with zero character overlap.
• Multi-hop reasoning becomes graph traversal: Once you have a clean knowledge graph, complex questions become simple path-finding problems that Claude can reason over.
• Start with Haiku for extraction, use Sonnet for resolution: This balances cost and quality — Haiku handles the volume, Sonnet handles the nuance.
• Always measure precision and recall: Build a small gold standard to validate your extraction quality before scaling to production.