CH-DDI
收藏资源简介:
CH-DDI是一个专门为捕获医学文本复杂语义而设计的中文药物-药物相互作用数据集。该数据集由郑州大学合作创新中心互联网医疗与中国医学科学院阜外医院共同创建,旨在支持大规模知识图谱构建及下游应用。数据集通过精心设计的滑动窗口策略进行跨度切分,并利用多头注意力和双向LSTM提取上下文语义信息,为实体识别和关系抽取任务提供支持。CH-DDI的构建考虑了关系重叠问题,即同一药物可能与多个其他药物具有相同或不同的相互作用,这对于准确的语义理解至关重要。该数据集的应用领域是中文医疗文本的实体和关系抽取,目的是为了促进大规模知识图谱的构建和完成各种下游任务。
CH-DDI is a Chinese drug-drug interaction (DDI) dataset specifically designed to capture complex semantics in medical texts. This dataset was co-developed by the Collaborative Innovation Center for Internet Healthcare of Zhengzhou University and Fuwai Hospital, Chinese Academy of Medical Sciences, aiming to support large-scale knowledge graph construction and downstream applications. The dataset conducts span segmentation via a meticulously designed sliding window strategy, and extracts contextual semantic information using multi-head attention and bidirectional LSTM, providing support for entity recognition and relation extraction tasks. The construction of CH-DDI considers the relation overlap problem, i.e., a single drug may have identical or distinct interactions with multiple other drugs, which is critical for accurate semantic understanding. The application scope of this dataset covers entity and relation extraction from Chinese medical texts, with the goal of facilitating the construction of large-scale knowledge graphs and completing various downstream tasks.

- 1The Joint Entity-Relation Extraction Model Based on Span and Interactive Fusion Representation for Chinese Medical Texts with Complex Semantics郑州大学计算机与人工智能学院, 合作创新中心互联网医疗, 中国医学科学院阜外医院, 首都医科大学附属北京友谊医院老年科 · 2025年



