基于本体的领域文档主题抽取方法研究

为了使抽取的主题词更能反映领域文档的内容,提出一种基于本体的领域文档主题抽取方法。该方法利用领域文档的特点,使用领域本体对文档词汇集进行过滤,排除非领域高频词汇的干扰并降低文档词汇集维度,从而提高算法效率和抽取质量;利用同/近义词典对文档候选主题词及其权重进行合并,降低同/近义词对抽取结果的影响,使得结果更加全面准确。实验表明,该方法具有较高的正确率和召回率。

Research Based on Domain Ontology Document Theme Extraction Method

In order to reflect contents of the extracted keywords field of the document, this paper proposes a field of document extraction method based on ontology. The method uses the characteristics of the field of document, which uses domain ontology to filter documentation vocabulary, to exclude the interference of high frequency vocabularies not in the field, and reduce the dimension of documentation vocabulary, thus improving the algorithm efficiency and extraction quality. This method uses synonym/near synonym dictionary to candidate theme words and their weights, reduces the impact of synonym/near synonym on extraction results, making the results more comprehensive and accurate. Experiments show that the method has higher precision and recall rate.