Biomedical Knowledge Graphs with Negative Statements

本数据集名为‘Biomedical Knowledge Graphs with Negative Statements’，由里斯本大学科学学院LASIGE创建。数据集包含三个子集，分别用于蛋白质-蛋白质相互作用预测、基因-疾病关联预测和疾病预测。这些数据集基于两个成功的生物医学本体：基因本体（GO）和人类表型本体（HP），并特别增加了否定声明以增强数据集的完整性和应用性能。数据集的创建过程涉及从生物信息数据库中提取实体对，并确保这些实体对具有正负声明。该数据集主要应用于生物医学领域，旨在通过考虑否定声明来提高知识图谱嵌入的性能，从而解决生物医学知识图谱中信息不完整的问题。

This dataset is named "Biomedical Knowledge Graphs with Negative Statements", and was created by LASIGE at the Faculty of Sciences, University of Lisbon. It comprises three subsets dedicated to protein-protein interaction prediction, gene-disease association prediction, and disease prediction respectively. These subsets are constructed upon two well-established biomedical ontologies: the Gene Ontology (GO) and the Human Phenotype Ontology (HP), with negative statements specially incorporated to enhance the dataset's completeness and application performance. The dataset development process involves extracting entity pairs from bioinformatics databases, and ensuring that these entity pairs have both positive and negative statements. Primarily utilized in the biomedical domain, this dataset aims to improve the performance of knowledge graph embedding by accounting for negative statements, thereby addressing the issue of incomplete information in biomedical knowledge graphs.