大规模开放数字商业知识图谱评测基准

数据集以开放的数字商业知识图谱OpenBG为基础构建，OpenBG是开放的数字商业知识图谱，是一个使用统一Schema组织、涵盖产品和消费需求的百万级多模态数据集。OpenBG由阿里巴巴藏经阁团队和浙江大学联合提供。OpenBG基础数据来源于阿里系产品（包含全球速卖通AliExpress）的真实流量日志。
OpenBG-IMG是电子商务领域的多模态知识图谱，来自于CCKS2022面向数字商务的知识处理与应用评测任务三：多模态商品知识图谱链接预测。OpenBG-IMG数据集相比原CCKS竞赛数据集，添加了新的dev划分，并且添加了实体和关系的文本信息，测试集来自CCKS竞赛初赛测试集。
OpenBG500是电子商务领域的知识图谱，包含500个关系，从OpenBG中筛选采样得到。OpenBG500-L是大规模电子商务领域知识图谱，包含500个关系（与OpenBG500一样）。
数据量为3.72GB

This dataset is built upon the open digital business knowledge graph OpenBG. OpenBG, an open digital business knowledge graph, is a million-scale multimodal dataset organized with a unified Schema, covering products and consumer demands. It is jointly provided by the Alibaba Scripture Pavilion Team and Zhejiang University. The underlying data of OpenBG is sourced from real traffic logs of Alibaba ecosystem products, including AliExpress.

OpenBG-IMG is an e-commerce domain multimodal knowledge graph, originating from Task 3 of the 2022 CCKS Evaluation on Knowledge Processing and Applications for Digital Commerce: Multimodal Commodity Knowledge Graph Link Prediction. Compared with the original CCKS competition dataset, the OpenBG-IMG dataset adds a new dev split, as well as textual information for entities and relations, while its test set is derived from the preliminary test set of the CCKS competition.

OpenBG500 is an e-commerce domain knowledge graph containing 500 relations, which is sampled and filtered from OpenBG. OpenBG500-L is a large-scale e-commerce knowledge graph that also contains 500 relations (identical to OpenBG500).

The total data volume is 3.72 GB.

搜集汇总

背景与挑战

背景概述

该数据集是基于阿里巴巴和浙江大学联合提供的OpenBG（开放数字商业知识图谱）构建的评测基准，涵盖电子商务领域的多模态商品知识图谱，包含OpenBG-IMG和OpenBG500等子集，数据来源于阿里系产品的真实流量日志，总数据量为3.72GB，用于支持数字商业知识处理与应用的相关评测任务。