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
import networkx as nx
Initialize clients
haiku_client = anthropic.Anthropic()  # For extraction
sonnet_client = anthropic.Anthropic()  # For resolution

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": fetch_wikipedia_summary("Moon"),
}

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.

class Entity(BaseModel):
    name: str = Field(description="The canonical name of the entity")
    type: str = Field(description="Entity type: PERSON, ORG, LOC, EVENT, or OTHER")
    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 in present tense")
    object: str = Field(description="Name of the object entity")
class Extraction(BaseModel):
    entities: List[Entity]
    relations: List[Relation]
def extract_from_document(text, title):
    response = haiku_client.messages.parse(
        model="claude-3-haiku-20240307",
        max_tokens=2000,
        system="You extract entities and relations from text. Be thorough but precise.",
        messages=[{
            "role": "user",
            "content": f"Extract all entities and their 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(text, title)

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(all_extractions):
    # Collect all unique entities with their descriptions
    entity_registry = {}
    for doc_title, extraction in all_extractions.items():
        for entity in extraction.entities:
            key = (entity.name, entity.type)
            if key not in entity_registry:
                entity_registry[key] = entity.description
    
    # Group by type for resolution
    from collections import defaultdict
    by_type = defaultdict(list)
    for (name, etype), desc in entity_registry.items():
        by_type[etype].append({"name": name, "description": desc})
    
    alias_to_canonical = {}
    for etype, entities in by_type.items():
        if len(entities) < 2:
            alias_to_canonical[entities[0]["name"]] = entities[0]["name"]
            continue
        
        prompt = f"""Group these {etype} entities that refer to the same real-world thing.
        For each group, choose the best canonical name.
        
        Entities:
        {chr(10).join(f'- {e["name"]}: {e["description"]}' for e in entities)}
        
        Return a JSON mapping from each original name to its canonical name."""
        
        response = sonnet_client.messages.create(
            model="claude-3-sonnet-20240229",
            max_tokens=1000,
            messages=[{"role": "user", "content": prompt}]
        )
        
        # Parse the mapping (simplified; use structured outputs in production)
        import json
        mapping = json.loads(response.content[0].text)
        alias_to_canonical.update(mapping)
    
    return alias_to_canonical
alias_map = resolve_entities(all_extractions)

Two failure modes to watch for. First, any raw name Claude leaves out of every cluster silently disappears from the graph, because alias_to_canonical has no entry for it — 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 traceability.

def build_graph(all_extractions, alias_map):
    G = nx.MultiDiGraph()
    
    for doc_title, extraction in all_extractions.items():
        # Add nodes with canonical names
        for entity in extraction.entities:
            canonical = alias_map.get(entity.name, entity.name)
            G.add_node(
                canonical,
                type=entity.type,
                source=doc_title,
                original_name=entity.name
            )
        
        # Add edges with canonical endpoints
        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 for the payoff: answering questions that require traversing multiple relations. We serialize the relevant subgraph back to Claude for reasoning.

def query_graph(graph, question):
    # Serialize the graph as a text representation
    nodes_info = "\n".join(
        f"- {node} ({data['type']})"
        for node, data in graph.nodes(data=True)
    )
    
    edges_info = "\n".join(
        f"- {u} --[{data['predicate']}]--> {v}"
        for u, v, data in graph.edges(data=True)
    )
    
    prompt = f"""Given this knowledge graph:
Nodes:
{nodes_info}
Edges:
{edges_info}
Answer this question using multi-hop reasoning through the graph:
{question}
Explain your reasoning step by step."""
    
    response = sonnet_client.messages.create(
        model="claude-3-sonnet-20240229",
        max_tokens=1000,
        messages=[{"role": "user", "content": prompt}]
    )
    
    return response.content[0].text
Example: multi-hop question
answer = query_graph(graph, "Which astronauts worked on projects that NASA funded?")
print(answer)

Measuring Extraction Quality

To trust your graph, you need to measure precision and recall against a gold standard. Create a small annotated set of documents with known entities and relations, then compare your extraction against it.

def evaluate_extraction(gold_entities, gold_relations, extracted_entities, extracted_relations):
    # Convert to sets for comparison
    gold_entity_set = set(gold_entities)
    extracted_entity_set = set(extracted_entities)
    
    # Precision: what we extracted that was correct
    true_positives = extracted_entity_set & gold_entity_set
    precision = len(true_positives) / len(extracted_entity_set) if extracted_entity_set else 0
    
    # Recall: what we missed
    recall = len(true_positives) / len(gold_entity_set) if gold_entity_set else 0
    
    return {
        "precision": precision,
        "recall": recall,
        "f1": 2  (precision  recall) / (precision + recall) if (precision + recall) > 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. A typical pipeline might use:

• Haiku for initial entity/relation extraction (cheap, fast, schema-constrained)
• Sonnet for entity resolution (needs to weigh conflicting evidence)
• Sonnet for multi-hop query answering (complex reasoning)

Monitor your token usage and adjust based on your accuracy requirements and budget.

Key Takeaways

• No training data needed: Claude's structured outputs let you extract entities and relations from any domain with just a prompt and a Pydantic model — no labeled data required.
• Entity resolution is critical: Raw extraction produces fractured graphs. Use Claude to collapse variants into canonical nodes, but watch for over-merging and silent drops.
• Multi-hop reasoning works: By serializing your knowledge graph back to Claude, you can answer questions that span multiple documents and relations without complex traversal logic.
• Measure what you build: Always evaluate precision and recall against a gold set. The cost/quality tradeoff between Haiku and Sonnet is real — choose based on your accuracy needs.
• Scale up easily: The in-memory approach transfers directly to production graph databases like Neo4j or Neptune when your data grows.