利用语义关系抽取生成生物医学文摘的算法

Automatic Summarization Algorithm for Biomedical Literature Based on Semantic Relation Extraction

通过自动摘要技术对生物医学概念进行摘要抽取,能够提高研究人员查阅和分析相关资料的效率。利用生物医学语义关系抽取多文档摘要,旨在从语义层面比较全面地覆盖查询概念的多方面内容,帮助研究人员快速掌握查询概念的主要信息。从生物医学文本中挖掘出了概念的重要语义关系,并利用语义关系作为衡量句子重要性的特征,生成查询概念的摘要。分析了H1N1、风湿病、脑脊髓炎等5种疾病,生成的摘要基本覆盖了这几种疾病的致病原因、类型、防治策略等语义类型。实验结果表明,利用语义关系特征抽取摘要的方法不但能提高摘要的性能,而且增加了生物医学语义层面内容,使生成的摘要更符合研究人员的查询需要。

Automatic summarization can help biomedical researchers to get a general idea of the given concept and make the research more efficient. Using semantic relation to extract summaries can cover information in more aspects on semantic level. Researchers can get the knowledge more easily. This paper extracts the important semantic relation of concepts from biomedical literatures, and uses the semantic relation as the character of measuring sen- tence importance to generate the summary. It focuses on five diseases, such as H1N1, rheumatism and encephalo- myelitis. Summary extracted contains the causes, types and treatments of the given diseases. Experimental results show that this method can improve the summarizing performance. Compared with the general method, the summarization with semantic relations can integrate the content of multi-document on semantic level and meet the need of biomedical researchers.