Objective Carrier air-hub operation parsing is a typical domain-specific structured understanding task in aircraft carrier scenarios. Its source data are mainly recorded in the form of unstructured mission logs, command messages, and support reports, and are characterized by strong temporal dependency, complex spatiotemporal constraints, dense domain terminology, and tight coupling among aircraft states, resource occupation, and operation stages. These characteristics make automatic parsing prone to erroneous outputs, omission of critical information, and structural inconsistency. Although cloud-based large language models offer a feasible solution owing to their broad prior knowledge and strong generative capability, their high computational and deployment costs limit their use in highly resource-constrained carrier-side environments. In contrast, locally deployable large models are more suitable for sensitive operational settings, yet their domain parsing capability remains insufficient. This study aims to construct a deployable domain-specific large model for carrier air-hub operation parsing and to improve its accuracy, reliability, and consistency under local deployment constraints.

Method We propose a domain-specific large model construction method for carrier air-hub operation parsing, termed the Carrier Air-Hub Operation Parsing Large Model (CAOP). The proposed framework consists of two closely coupled stages, namely high-dimensional distillation during training and collaborative verification during inference. In the training stage, a cloud-based teacher model is used to provide structured supervision across six key information dimensions, including entity recognition, relation recognition, statistical information extraction, predictive information extraction, text and numerical joint semantic parsing, and abnormal event cue extraction. These dimension-specific outputs are further aligned and fused to construct transferable supervision signals, through which the domain parsing capability of the teacher model is injected into a locally deployable student model by parameter-efficient optimization. In the inference stage, a carrier aviation operation knowledge base and case retrieval mechanism are introduced to provide external evidence and similar historical references. On this basis, collaborative verification is performed on candidate outputs by imposing evidence constraints and structural consistency constraints, so as to suppress hallucination, reduce redundancy, and correct inconsistent fields.

Results Under the experimental setting adopted in this study, the proposed method improves the average score of the local Qwen3-32B model on carrier air-hub operation parsing from 13.3 to 18.4, with an absolute gain of 5.1 points and a relative improvement of approximately 38.3%. The resulting performance is higher than that of the compared cloud-based models and is close to expert assessment. The gains are mainly reflected in more complete field extraction, stronger evidence correspondence, and better structural consistency of the final parsing results.

Conclusion The results demonstrate that CAOP significantly enhances the accuracy, reliability, and consistency of carrier air-hub operation parsing under local deployment constraints. It provides a reliable data foundation and trustworthy structured inputs for subsequent tasks including automatic scheduling plan generation, conflict detection, and optimization solving.