海关企业/商品知识图谱数据集

本数据集是项目组研究人员依据海关实际业务需求、围绕企业与商品主题进行设计、基于海关业务数据生成的知识图谱数据集。本数据集以海关的业务数据如报关单、企业基本信息、企业处罚信息、货物查验信息等为原始数据源，进行深度数据清洗、实体识别、关系抽取后按照设定的结构进行生成，包含企业、商品、税号、物流工具等类型的实体与相关关系、属性的数据，同时经过脱敏处理，由图数据库导出为367MB的CSV数据文件。本数据集包含118万个企业实体、12万个商品实体以及170万条关系，支持海关企业、商品节点与关系的查询，并可通过图论算法、图卷积算法等人工智能技术应用于多类风险分析，如社群挖掘、风险转移等，为海关监管决策与风险防控提供智能化支撑。

This dataset is a knowledge graph developed by project researchers, which is designed based on actual customs business needs, centered on the themes of enterprises and commodities, and generated from customs operational data. Taking customs business data including customs declarations, enterprise basic information, enterprise penalty records, goods inspection records and other relevant materials as the original data source, this dataset is produced after in-depth data cleaning, entity recognition and relation extraction, following a pre-defined structure. It contains data of entities such as enterprises, commodities, tax codes, logistics tools and other types, along with their associated relationships and attributes, and has undergone data desensitization. The dataset is exported as a 367MB CSV file from a graph database. This dataset comprises 1.18 million enterprise entities, 0.12 million commodity entities and 1.7 million relational entries. It supports queries for customs enterprise and commodity nodes as well as their corresponding relationships, and can be applied to multiple risk analysis scenarios via artificial intelligence technologies like graph theory algorithms and graph convolutional algorithms, including community mining and risk transfer, providing intelligent support for customs supervision decision-making and risk prevention and control.

搜集汇总

背景与挑战

背景概述

该数据集是基于海关业务数据生成的知识图谱数据集，包含企业、商品、税号等实体及关系，经过脱敏处理，数据量为364.22MB，包含118万企业实体和12万商品实体，支持海关风险分析应用。