A2H_Linking_Data_MONDO_UMLS

The `data` folder contains the ontology resources used for entity linking. For each terminology (`mondo` and `umls`), it provides: - `embeddings/` — precomputed ontology term embeddings- `*_term_id_pairs.json` or `*_term_id_pairs_combined.json` — mappings between surface forms / synonyms and ontology IDs- `*_id_to_term_map.json` — mappings from ontology IDs to canonical normalized terms These resources are used during SapBERT-based normalization of extracted drug and disease mentions. Expected structure: data/├── mondo/│ ├── embeddings/│ ├── mondo_term_id_pairs.json│ └── mondo_id_to_term_map.json└── umls/ ├── embeddings/ ├── umls_term_id_pairs_combined.json └── umls_id_to_term_map.json The ontology resources were generated in the upstream **Preclinical_Pipeline** project and are not created within this repository itself. Generation code and documentation are available here: - https://github.com/Ineichen-Group/Preclinical_Pipeline Specifically, these files are produced in /04_normalization, see docu in the main README under Named entity normalization (NEN).`data` 文件夹包含用于实体链接的本体资源。针对 `mondo` 和 `umls` 两个术语集，分别提供以下内容： - `embeddings/`：预计算得到的本体术语嵌入向量 - `*_term_id_pairs.json` 或 `*_term_id_pairs_combined.json`：表层形式/同义词与本体ID的映射表 - `*_id_to_term_map.json`：本体ID至标准归一化术语的映射表 上述资源将在基于SapBERT的提取药物与疾病提及的归一化流程中被使用。 预期目录结构如下： data/ ├── mondo/ │ ├── embeddings/ │ ├── mondo_term_id_pairs.json │ └── mondo_id_to_term_map.json └── umls/ ├── embeddings/ ├── umls_term_id_pairs_combined.json └── umls_id_to_term_map.json 本仓库并未生成这些本体资源，它们均源自上游的 **Preclinical_Pipeline** 项目。该项目的生成代码与文档可通过以下链接获取： - https://github.com/Ineichen-Group/Preclinical_Pipeline 具体而言，这些文件生成于该项目的 `/04_normalization` 目录，相关说明可查看主README文档中的「命名实体归一化（Named Entity Normalization, NEN）」章节。