论文部分内容阅读
【目的】利用文献的主题标引结果,发现其中隐含的重要语义关系。【方法】基于MEDLINE数据库中的生物医学主题标引文献,提出一种语义关系发现算法,涉及主题词组配原则、主题标引规则以及基于加权标引词和关系出现频次的优化方法等多个环节。【结果】收集疾病与症状方面的实验数据对算法进行实验验证,并结合领域专家审核,结果表明本文所发现语义关系的准确率可达到95%以上。【局限】本文所研究的语义关系发现算法仅适用于具有主题标引结果的文献。【结论】从大规模生物医学主题标引文献中发现中英文两种语言的语义关系是有效可行的,对其他领域语义关系的发现具有极高的借鉴意义。
【Objective】 Using the indexing results of the literature, we found the important semantic relations implied in it. 【Method】 Based on the biomedical subject indexing documents in the MEDLINE database, a semantic relationship discovery algorithm was proposed, which involved the keyword assignment principle, the subject indexing rules and the optimization method based on the weighted indexing words and the relation frequency . 【Results】 The experimental data on the collection of diseases and symptoms were validated experimentally and verified by field experts. The results showed that the accuracy of the semantic relations found in this paper can reach more than 95%. [Limitations] The semantic relation discovery algorithm studied in this paper is only applicable to documents with thematic indexing results. 【Conclusion】 It is feasible and effective to find the semantic relationship between Chinese and English languages from large-scale biomedical subject indexing documents. It is of great reference to the discovery of semantic relations in other fields.