Skip to main content
Back to challenges
Challenge

Industrial Manual Knowledge Graph & Agent RAG using LlamaIndex

Factories accumulate thousands of unorganized PDF operating manuals, Rosemount sensor technical specs, and historical work orders. Simple vector search often returns irrelevant snippets when engineers query complex multi-part component specifications. In this challenge, developers will utilize LlamaIndex to build an advanced RAG and Knowledge Graph engine over industrial documentation. The system extracts structured node relationships (e.g., [Sensor] --MEASURES--> [pH] --REQUIRES--> [Calibration Buffer Solutions]) and combines Knowledge Graph Index with VectorStoreIndex query engines in LlamaIndex. When queried by control engineers, the agent synthesizes pinpoint operational solutions with clear source citations.

EngineeringHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Build an advanced RAG and Knowledge Graph query engine using LlamaIndex over industrial equipment manuals and sensor specifications.

Delivery guide

How work is evaluated

Evaluation

Evaluates answer accuracy and citation source precision on technical queries against Emerson Rosemount manuals.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Buffer Specification Accuracy

Extracted pH buffers must accurately contain 4.01, 7.00, and 10.01

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

Citation Recall Rate

Percentage of returned answers containing exact valid document page references • target: 0.95 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Construct a LlamaIndex KnowledgeGraphIndex from unstructured technical PDFs

  • Combine VectorStoreIndex and PropertyGraphIndex using LlamaIndex RouterQueryEngine

  • Implement metadata filtering based on sensor model numbers (e.g., Rosemount 350A)

  • Generate precise technical troubleshooting answers with exact document citations

Resources and assets

Reference links and supporting material

Dataset notes

Unstructured text extracts and PDF documentation for industrial pH/ORP sensors, flowmeters, and SCADA gateway controllers.

Challenge setup

Evaluates answer accuracy and citation source precision on technical queries against Emerson Rosemount manuals.

Challenge input

Query string (e.g., 'What is the required calibration buffer pH for Rosemount 350A?')

LlamaIndex

Assigned primary orchestration framework

Evaluated output

JSON object with answer_text, citation_document, extracted_ph_buffers (array)

Evaluation checks
  • Extracted pH buffers must accurately contain 4.01, 7.00, and 10.01
  • Percentage of returned answers containing exact valid document page references • target: 0.95 • range: 0-1
Configured evaluation evidence
  • Benchmark: Enterprise Document QA Benchmark
  • Citation Recall Rate target: 95%
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Llama Index · Required
  • LlamaIndex · Optional
  • Llama 3.2 · Optional

Evaluation contract

  • Buffer Specification Accuracy · Weight 1
  • Citation Recall Rate · Weight 1

Metrics

  • Citation Recall Rate · Target 0.95

Benchmarks

  • Enterprise Document QA Benchmark
  • general

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: Query string (e.g., 'What is the required calibration buffer pH for Rosemount 350A?')

    Output: JSON object with answer_text, citation_document, extracted_ph_buffers (array)

    Execution harness: Python

Evaluation tools

  • LlamaIndex · Data framework & RAG engine

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Industrial Manual Knowledge Graph & Agent RAG using LlamaIndex