Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Compliance-Preserving Retrieval System for Aircraft MRO Task Search

Byungho Jo

TL;DR

The paper tackles a critical bottleneck in aircraft MRO by designing a compliance-preserving retrieval system that leverages revision-robust ATA metadata and vision-language structuring to map natural-language queries to exact certified procedures without modifying immutable manuals. It introduces a two-stage architecture with offline knowledge structuring and an online, LLM-assisted re-ranking pipeline that outputs JSON indices to preserve auditability. Across a large synthetic benchmark (>49k queries) and a bilingual study with 10 AMTs, the approach achieves over 90% retrieval accuracy and reduces lookup time by more than 95%, demonstrating practical applicability in multilingual, regulatory settings. These results show that semantic retrieval can significantly improve operational efficiency in aviation MRO while maintaining strict regulatory and traceability constraints.

Abstract

Aircraft Maintenance Technicians (AMTs) spend up to 30% of work time searching manuals, a documented efficiency bottleneck in MRO operations where every procedure must be traceable to certified sources. We present a compliance-preserving retrieval system that adapts LLM reranking and semantic search to aviation MRO environments by operating alongside, rather than replacing, certified legacy viewers. The system constructs revision-robust embeddings from ATA chapter hierarchies and uses vision-language parsing to structure certified content, allowing technicians to preview ranked tasks and access verified procedures in existing viewers. Evaluation on 49k synthetic queries achieves >90% retrieval accuracy, while bilingual controlled studies with 10 licensed AMTs demonstrate 90.9% top-10 success rate and 95% reduction in lookup time, from 6-15 minutes to 18 seconds per task. These gains provide concrete evidence that semantic retrieval can operate within strict regulatory constraints and meaningfully reduce operational workload in real-world multilingual MRO workflows.

A Compliance-Preserving Retrieval System for Aircraft MRO Task Search

TL;DR

The paper tackles a critical bottleneck in aircraft MRO by designing a compliance-preserving retrieval system that leverages revision-robust ATA metadata and vision-language structuring to map natural-language queries to exact certified procedures without modifying immutable manuals. It introduces a two-stage architecture with offline knowledge structuring and an online, LLM-assisted re-ranking pipeline that outputs JSON indices to preserve auditability. Across a large synthetic benchmark (>49k queries) and a bilingual study with 10 AMTs, the approach achieves over 90% retrieval accuracy and reduces lookup time by more than 95%, demonstrating practical applicability in multilingual, regulatory settings. These results show that semantic retrieval can significantly improve operational efficiency in aviation MRO while maintaining strict regulatory and traceability constraints.

Abstract

Aircraft Maintenance Technicians (AMTs) spend up to 30% of work time searching manuals, a documented efficiency bottleneck in MRO operations where every procedure must be traceable to certified sources. We present a compliance-preserving retrieval system that adapts LLM reranking and semantic search to aviation MRO environments by operating alongside, rather than replacing, certified legacy viewers. The system constructs revision-robust embeddings from ATA chapter hierarchies and uses vision-language parsing to structure certified content, allowing technicians to preview ranked tasks and access verified procedures in existing viewers. Evaluation on 49k synthetic queries achieves >90% retrieval accuracy, while bilingual controlled studies with 10 licensed AMTs demonstrate 90.9% top-10 success rate and 95% reduction in lookup time, from 6-15 minutes to 18 seconds per task. These gains provide concrete evidence that semantic retrieval can operate within strict regulatory constraints and meaningfully reduce operational workload in real-world multilingual MRO workflows.

Paper Structure

This paper contains 13 sections, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Artificial Intelligence in Aircraft MRO
  4. Large Language Models in Aviation
  5. Proposed System Architecture
  6. Offline Maintenance Task Knowledge Structuring
  7. Manual Retrieval Pipeline Architecture
  8. Experiments
  9. Synthetic Benchmark
  10. Human Study Design
  11. Performance Analysis
  12. Knowledge Structuring Quality
  13. Conclusion

Figures (2)

  • Figure 1: ATA chapter-based manual structure illustrating the hierarchical complexity that AMTs must navigate to locate specific maintenance tasks. A representative example, Ch.32 $\rightarrow$ 32-41 $\rightarrow$ 32-41-31 $\rightarrow$ 401 $\rightarrow$ 32-41-41-000-801, demonstrates a five-level navigation path with over fifty branching options. Numbers in blue indicate the task counts at each level.
  • Figure 2: Offline workflow extracts and structures tasks from maintenance PDF manuals into a Task Knowledge DB. During the online workflow, a technician query triggers Top-K retrieval with LLM re-ranking, previews the ranked tasks, and opens the certified procedure in the official viewer, maintaining full compliance while reducing lookup time from minutes to seconds.