Computational Notebook - Methods applied to the hoard of Le Câtillon II in the project ClaReNet

ClaReNet is a joint project of the Römisch-Germanische Kommission (German Archaeological Institute) and the Big Data Lab (Goethe University Frankfurt), funded by the German Federal Ministry of Education and Research (BMBF). It tests the possibilities and limits of new digital methods of classification and representation. This supplement provides a snapshot of the methods used in the project, which will be published in the paper "Supporting the analysis of a large coin hoard with AI-based methods'' submitted to CAA 2023, and is based on the Github repository "https://github.com/Frankfurt-BigDataLab/2023_CAA_ClaReNet".

The core of the analysis were digital images of around 70,000 coins discovered in a coin hoard at Le Câtillon II on the island of Jersey (UK), and the methods used fall into four parts:
1. Object detection: cropping the image to the coin and reducing bias;
2. Unsupervised learning: observing the way the coins are grouped independently of domain information;
3. Supervised learning: combining the results of unsupervised learning and domain information;
4. Image matching: detecting similar dies.
The snapshot is used to show and reproduce the analysis at the time of the paper. Further information and sources can be found on the official Github repository.

ClaReNet是罗马-日耳曼委员会（Römisch-Germanische Kommission，德国考古研究所）与大数据实验室（Big Data Lab，法兰克福歌德大学（Goethe University Frankfurt））的联合项目，由德国联邦教育与研究部（BMBF）资助。该项目旨在探究新型数字化分类与展示方法的应用潜力与边界。本补充材料呈现了项目所采用方法的简要快照，相关内容将收录于提交至CAA 2023的论文《基于人工智能方法辅助大型钱币窖藏分析》，本补充材料基于GitHub仓库"https://github.com/Frankfurt-BigDataLab/2023_CAA_ClaReNet"构建。

本次分析的核心数据集为英国泽西岛勒卡蒂永II号遗址出土的一处钱币窖藏中约7万枚钱币的数字影像，项目所采用的分析方法分为四个部分：
1. 目标检测（Object detection）：将图像裁切至钱币区域，降低数据偏差；
2. 无监督学习（Unsupervised learning）：不依赖专业领域知识，自主探索钱币的分组模式；
3. 监督学习（Supervised learning）：融合无监督学习结果与考古领域专业信息；
4. 图像匹配（Image matching）：检测相似币模。

本快照用于展示并复现论文撰写时的完整分析流程，更多相关信息与数据源可访问官方GitHub仓库获取。