Your Parts Book Is Already a Graph

A parts book — assemblies, positions, parts, machines — is a graph whether or not you model it as one. This article shows the nodes and edges hiding inside every parts book and what becomes possible when you make them explicit.

A parts book is a collection of assemblies, positions, parts and machines connected by relationships: an assembly contains positions, a position maps to a part, a part fits a machine, a part is replaced by a successor. Those are nodes and edges — a graph — whether or not you model them that way.

Open any parts book. Pick an assembly. Find position 14. Note the part number. Now ask: what other assemblies use this same part? Which machines does it fit? Has it been superseded? What replaced it? If the replacement has a different pack size, how many do I need?

Each of those questions is a graph traversal. The PDF cannot answer any of them. A flat spreadsheet can answer the first one, slowly. A relational database can answer most of them with enough joins. A graph answers all of them by following edges.

The nodes

Every parts book contains four types of entities:

Assembly — a group of components that form a functional unit. In the parts book, an assembly is typically one exploded-view page: the hydraulic system, the engine block, the control panel.

Position — a slot in an assembly where a part goes. Position 14 in the hydraulic assembly is a specific location, identified by a callout number on the exploded view. The position is stable: it does not change when the part at that position changes.

Part — a product identity. Not just a part number, but the resolved entity behind it: the current number, the supersession history, the cross-references, the specifications. A single part may appear in many positions across many assemblies.

Machine — a piece of equipment identified by make, model and serial range. Machines connect to assemblies through applicability data.

The edges

The relationships between these nodes are the edges that make the graph useful:

Edge	From	To	Meaning
CONTAINS_POSITION	Assembly	Position	This assembly has a slot at this position
MAPS_TO	Position	Part	The part installed at this position (scoped by applicability)
REPLACED_BY	Part	Part	The manufacturer has superseded this part with a new one
EQUIVALENT_TO	Part	Part	These two parts are cross-referenced as equivalents (from different brands)
FITS	Part (via Assembly)	Machine	This assembly — and therefore its parts — applies to this machine in this serial range
COMPONENT_OF	Part	Part	This part is a component of a larger assembly or kit

What the graph makes possible

Forward traversal: supersession resolution

A technician searches for part HY-3200. The graph follows the REPLACED_BY edge to HY-3450, then checks whether HY-3450 has its own REPLACED_BY edge (it does not — it is the current part). The result: “HY-3200 was superseded by HY-3450 on 2024-03-01. Source: manufacturer bulletin #412.”

A PDF cannot do this. A flat table can do it for one hop. The graph does it for arbitrarily deep chains, with the date and source on every link.

Reverse traversal: where-used

Given part HY-3450, the graph traverses incoming MAPS_TO edges to find every position that uses it, then follows CONTAINS_POSITION edges to find every assembly, then follows FITS edges to find every machine. The result: “HY-3450 is used in position 14 of the hydraulic assembly, which applies to model XR-700 (serial 42000 onwards) and model XR-800 (all serials).”

This is the “where-used” query that service teams need every time a part is discontinued, recalled, or back-ordered. In a PDF, it requires searching every page of every parts book. In the graph, it is one traversal.

Impact analysis: supersession ripple

When the manufacturer supersedes HY-3450 with HY-3500, the graph immediately shows the impact: every position that maps to HY-3450, every assembly that contains those positions, every machine that those assemblies apply to. The catalog team knows which pages, which BOMs, and which customer-facing listings need to be updated.

Cross-reference through shared assemblies

Two parts from different brands appear in the same position of equivalent assemblies for different machines. The graph captures this: Brand-A part BA-3200 and Brand-B part BB-3200 both map to position 14 of a hydraulic pump assembly. This positional evidence — two parts interchangeable in the same slot — is stronger than a dimension-only cross-reference.

What a PDF flattens

A PDF parts book contains all of these nodes and edges, but flattened:

Assemblies are pages
Positions are callout numbers printed on an image
Parts are text strings in a table
Machines are a header note
Edges are implicit (the reader infers them by proximity on the page)

The information is there. The structure is not. Making the structure explicit — turning pages into nodes and proximity into edges — is what separates a document from a data asset.

Where Claro fits

Claro provides the identity and relationship layer that makes the parts-book graph live. Each position points to a product identity that follows supersession chains, carries cross-references with evidence, and connects to applicability data by serial range. The exploded view stays visual. The graph behind it powers the traversals that the PDF never could.

Glossary

Exploded view, parts list and BOM

What each artifact carries and where they diverge.

Guide

Parts book data model

The four layers and why separating them keeps the book current.

Playbook

Convert a PDF parts book

Extraction and structuring workflow.

Glossary

Position and callout numbers

The reference numbers linking diagram to table.

FAQ

Why is a parts book a graph?

Because a parts book contains entities (assemblies, positions, parts, machines) and relationships between them (assembly contains position, position maps to part, part fits machine, part replaced by successor). Those are nodes and edges. A PDF flattens them into pages; a relational database normalizes them into tables; but the underlying structure is a graph.

What does modeling a parts book as a graph make possible?

Traversal: follow a supersession chain from an old part to the current one. Reverse lookup: given a part, find every assembly and machine it appears in. Impact analysis: when a part is superseded, find every position and every machine affected. Cross-reference: link parts across brands through the assemblies they share. None of these are practical with a PDF or a flat table.