地球资源数据云——数据资源详情
合作专利分类代码含义
该数据集《Cooperative Patent Classification Codes Meaning》主要用于多分类任务，数据形态以图像为主，应用场景偏向金融风控。 题目说明：For the US Patent competition. Source: cooperativepatentclassification.org 任务类型：图像多分类。 建议流程：先检查类别分布与脏样本，再用迁移学习（如 ResNet/EfficientNet）建立基线。 评估建议：使用分层切分或交叉验证，优先关注 F1、Recall、AUC 等分类指标。 可用文件：titles.csv。 Context This dataset is provided for the U.S. Patent Phrase to Phrase Matching competition. It adds additional information by providing the meaning of each code in the context column. For more info, check out the discussion thread, and take a look at this starter notebook to see how to incorporate the data into the competition. Content Preprocessing script here: https://www.kaggle.com/code/xhlulu/download - and - process - cpc
摘要概览
该数据集《Cooperative Patent Classification Codes Meaning》主要用于多分类任务，数据形态以图像为主，应用场景偏向金融风控。 题目说明：For the US Patent competition. Source: cooperativepatentclassification.org
任务类型：图像多分类。
建议流程：先检查类别分布与脏样本，再用迁移学习（如 ResNet/EfficientNet）建立基线。
评估建议：使用分层切分或交叉验证，优先关注 F1、Recall、AUC 等分类指标。
可用文件：titles.csv。
Context
This dataset is provided for the U.S. Patent Phrase to Phrase Matching competition. It adds additional information by providing the meaning of each code in the context column.
For more info, check out the discussion thread, and take a look at this starter notebook to see how to incorporate the data into the competition.
Content
Preprocessing script here: https://www.kaggle.com/code/xhlulu/download - and - process - cpc