基于大语言模型的铁路多式联运货物品名智能分类识别与映射OA
Cargo Names Intelligent Classification Identification and Mapping for Railway Multimodal Transport Based on Large Language Models
为解决铁路多式联运中电商货物品名快速识别与标准化归类问题,提出一种基于大语言模型的货物品名智能分类识别与映射技术方案和系统实现.以Qwen2.5-14B为基座模型,通过本地部署及面向领域微调,实现从冗余信息较多的电商品名中提取本体词,采用向量相似度检索获取铁路货物品名候选集,再结合语义推理与融合置信度决策机制,完成向铁路品名字典的自动化映射.应用结果表明,该方法能有效提升货物品名分类映射的准确性与处理效率,具备良好的适应性与持续优化能力,能够满足铁路多式联运的实际业务需求,显著提升了物流作业效率与企业智能化水平.
To address the challenges of rapid identification and standardized classification of e-commerce cargo names in railway multimodal transport,this paper proposed a technical solution and system implementation for intelligent classification recognition and mapping of cargo names based on large language models.Using Qwen2.5-14B as the base model,the extraction of ontology terms from e-commerce cargo names with rich redundant information through local deployment and domain-specific fine-tuning.Vector similarity retrieval was employed to obtain candidate sets of railway cargo names.Through a semantic reasoning and confidence-based decision-making mechanism,automated mapping to the railway cargo name dictionary was achieved.Application results demonstrate that the proposed method effectively improves the accuracy and processing efficiency of cargo name classification mapping,with strong adaptability and continuous optimization capability,which meets the practical business requirements of railway multimodal transport,and significantly enhances logistics operation efficiency and enterprise intelligent level.
冯轶斌
中铁集装箱运输有限责任公司 多式联运事业部,北京 100073
交通工程
多式联运货物品名本体词大语言模型智能映射
Multimodal TransportCargo NameOntology TermLarge Language ModelIntelligent Mapping
《铁路物流》 2026 (9)
64-70,7
中铁国际多式联运有限公司科研项目(DLHT-TH-2025-00328)
评论