DeepLobe：2021年面向机器学习与深度学习项目的数据标注权威指南（英文版）

The AI life cycle begins with collecting the data and organizing it. Data labeling - also synonymous with data annotation/tagging/classification in many scenarios - is the process of identifying raw data and giving meaningful and informative labels to this data to provide some context for the machine learning models to learn from it. Curating the data manually by adding keywords to the unstructured data enables the machines to automatically recognize the concepts that these keywords describe.