arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.09231cs.AI
收藏

认知分层数据合成用于LLM在空间态势感知领域的领域适应

Cognitively Layered Data Synthesis for Domain Adaptation of LLMs to Space Situational Awareness

Qian Xuesen Laboratory of Space Technology(钱学森空间技术实验室);China Academy of Space Technology(中国航天科技研究院);State Key Laboratory of Space Information System and Integrated Application(空间信息系统与集成应用国家重点实验室)

Ding Linghu, Cheng Wang, Da Fan, Wei Shi, Kaifeng Yin, Xiaoliang Xue, Fan Yang, Haiyi Ren, Cong Zhang

AI总结:

本文提出BD-FDG框架,通过认知分层问题建模和结构化知识组织,构建高质量SFT数据集,提升LLM在空间态势感知领域的适应性能。

摘要:

大型语言模型(LLMs)在通用任务上表现出色。然而,将其转移到复杂的工程领域,如空间态势感知(SSA),仍然具有挑战性,原因在于结构对齐不足,缺乏更高阶的认知监督,以及数据质量标准与工程规范之间的对应关系不佳。核心瓶颈是高质量监督微调(SFT)数据集的构建。为此,我们提出了BD-FDG(基于布洛姆分类法的领域特定微调数据生成)框架,通过三种机制解决知识覆盖不全、认知深度浅和质量可控性有限的问题:结构化知识组织、认知分层问题建模和自动化质量控制。该框架使用知识树确保语料库的结构覆盖,设计一个跨越九个类别和六个认知层次(从记忆到创造)的问题生成方案,生成具有连续难度梯度的样本,并应用多维评分流程以确保领域严谨性和一致性。使用BD-FDG,我们构建了SSA-SFT,一个约23万样本的领域数据集,并对Qwen3-8B进行微调以获得SSA-LLM-8B。实验表明,SSA-LLM-8B在领域测试集上实现了相对BLEU-1提升144%(无思考)和176%(有思考),在竞技比较中胜率高达82.21%,同时在通用基准测试(MMLU-Pro,MATH-500)中保持了大致表现。这些结果验证了以认知分层驱动的SFT数据构建作为复杂工程领域有效范式,并提供了可转移的领域特定LLM适应框架。

英文摘要:

Large language models (LLMs) demonstrate exceptional performance on general-purpose tasks. however, transferring them to complex engineering domains such as space situational awareness (SSA) remains challenging owing to insufficient structural alignment with mission chains, the absence of higher-order cognitive supervision, and poor correspondence between data quality criteria and engineering specifications. The core bottleneck is the construction of high-quality supervised fine-tuning (SFT) datasets. To this end, we propose BD-FDG (Bloom's Taxonomy-based Domain-specific Fine-tuning Data Generation), a framework that addresses incomplete knowledge coverage, shallow cognitive depth, and limited quality controllability through three mechanisms: structured knowledge organization, cognitively layered question modeling, and automated quality control. The framework uses a knowledge tree to ensure structured corpus coverage, designs a question generation scheme spanning nine categories and six cognitive levels from Remember to Create to produce samples with a continuous difficulty gradient, and applies a multidimensional scoring pipeline to enforce domain rigor and consistency. Using BD-FDG, we construct SSA-SFT, a domain dataset of approximately 230K samples, and fine-tune Qwen3-8B to obtain SSA-LLM-8B. Experiments show that SSA-LLM-8B achieves relative BLEU-1 improvements of 144\% (no-think) and 176\% (think) on the domain test set and a win rate of 82.21\% over the baseline in arena comparisons, while largely preserving general benchmark performance (MMLU-Pro, MATH-500). These results validate SFT data construction driven by cognitive layering as an effective paradigm for complex engineering domains and provide a transferable framework for domain-specific LLM adaptation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑