Case Study: Caterpillar achieves faster, more efficient equipment repairs with Neo4j-powered NLP

Neo4j Provides Natural Language Processing at Scale, Making Equipment Repair More Efficient

Caterpillar, the world’s leading maker of construction and mining equipment, faced a big data challenge: more than 27 million technician-generated repair and warranty documents contained valuable but dispersed information. Relational systems returned unparsed text that couldn’t reveal the underlying causes, trends or actionable insights needed to speed repairs and improve maintenance decisions.

Caterpillar built an NLP pipeline around Neo4j’s graph database—ingesting text via Python and open-source NLP tools, using WordNet and the Stanford Dependency Parser, and applying a machine-learning classifier to tag and link concepts into ontologies. The graph-based solution enables real-time searches across millions of documents, uncovers cause-and-effect relationships, prescribes next steps for faults and materially increases the efficiency and accuracy of equipment repair and maintenance.

Caterpillar

Ryan Chandler

Chief Data Scientist