Bio-ML track at OAEI 2022

本研究基于Mondo和UMLS创建了五个新的生物医学OM任务，旨在解决现有Ontology Matching (OM)系统评估的局限性。数据集包括从Mondo和UMLS提取的ontology，涉及疾病和医学概念，通过人工校验确保参考映射的质量。创建过程中采用了ontology修剪技术，以提高映射的相对完整性，并生成不同大小的ontology以适应不同计算特性的OM系统。数据集适用于ML和非ML基础的OM系统，通过统一的评估框架，包括MRR和Hits@K等排名指标，以及Precision、Recall和F-score等全局匹配指标，全面评估OM系统性能。这些资源将用于OAEI 2022的新Bio-ML赛道，特别吸引ML基础的OM系统参与，以推动OM领域的研究和发展。

This study developed five novel biomedical Ontology Matching (OM) tasks based on Mondo and UMLS, aiming to address the limitations of existing ontology matching system evaluations. The dataset comprises ontologies extracted from Mondo and UMLS, covering disease and medical concepts, with manual validation conducted to ensure the quality of reference alignments. Ontology pruning techniques were adopted during the creation process to enhance the relative completeness of alignments, and ontologies of varying sizes were generated to accommodate OM systems with different computational characteristics. This dataset is applicable to both ML-based and non-ML-based OM systems, enabling comprehensive evaluation of OM system performance via a unified evaluation framework, which includes ranking metrics such as Mean Reciprocal Rank (MRR) and Hits@K, as well as global matching metrics including Precision, Recall, and F-score. These resources will be used for the new Bio-ML track of OAEI 2022, which specifically invites ML-based OM systems to participate, so as to promote research and development in the ontology matching field.