研究现有术语语义层次关联的主要方法,针对如何自动获取并且准确描述术语语义层次关系这一关键问题,提出构建的术语语义层次获取流程,主要基于确定性退火的多重聚类算法获取层次关系。最后选取了部分中文专利文本摘要数据集对流程的效果进行测试与评估,取得了不错的实验结果。
Term semantic relationship is a key step of Chinese text information processing.Through researching on some exis-ting methods,this paper proposed a process of term semantic hierarchy induction.It mainly used multiple clustering method based on deterministic annealing to get the whole hirarchy.Finally,it selected some Chinese patent data to test and evaluate the effectiveness of the process and it made well.