基于本体的林业领域文档特征权重模型-计算机工程与应用.PDF

20 2013 ，49 （18） Computer Engineering and Applications 计算机工程与应用 基于本体的林业领域文档特征权重模型 张乃静，鞠洪波，纪 平 ZHANG Naijing, JU Hongbo, JI Ping 中国林业科学研究院 资源信息研究所，北京 100091 Research Institute of Forestry Information Techniques, Chinese Academy of Forestry, Beijing 100091, China ZHANG Naijing, JU Hongbo, JI Ping. Modeling feature weight of document of forestry domain based on ontology. Com- puter Engineering and Applications, 2013, 49 （18）：20-23. Abstract ：In the traditional feature weight of documents calculating, the model only considers the key word but other more relative words, so that the results of information retrieval are not comprehensive and precise. Aiming to solve these disadvantages above, this paper presents a model that calculates feature weight of document of forestry domain based on ontology. The steps of this model are as follows: calculate the feature weight using TF-IDF model; require the semantic distance, contact ratio and level difference between the key word and other relative words of document based on ontology, and then calculate the semantic simi- larity; calculate the feature weight using both results of TF-IDF and semantic similarity. The experiment proves that this improved model can increase the precision and recall ratio in documents retrieval, and meets the needs of users satisfactorily. Key words ：ontology; forestry domain; document feature; ranking model; semantic similarity 摘 要：传统文档特征权重模型仅考虑关键词本身，文档内其他相关词汇并没有参与计算，信息检索时无法返回全面和准确 的结果。为解决该问题提出了一种基于本体的林业领域文档特征权重模型。该模型计算TF-IDF 特征权重；结合林业领域本 体，分别获取关键词和林业领域内其他词汇的语义距离、语义重合度和概念的层次差，并计算语义相关度；结合TF-IDF 和语 义相似度的结果计算特征权重。实验证明该模型可以提高文本检索的查准率和查全率，使检索结果更加满足用户的需求。 关键词：本体；林业领域；文档特征；权重模型；语义相似度 文献标志码：A 中图分类号：TP391 doi ：10.3778/j.issn. 1002-8331.1303-0173 1 引言 和查准率；文献[5]利用本体知识库推理实现了语义搜索； 现有林业领域信息检索方法多是以关键词匹配为基 文献[6]利用本体改进了向量空间模型中排名算法；文献[7] 础的检索方式，随着林业相关研究的深入，林业领域信息 利用本体中概念的语义距离来计