基于BMRM迭代排序方法的本体学习算法

本体作为一种结构化数据存储和表示模型已成为信息检索领域的研究热点,并被应用于生物医学、地理科学、社会科学等诸多领域。提出基于BMRM迭代排序学习方法的本体相似度计算和本体映射算法,利用BMRM迭代得到最优参数向量,由此得到排序函数,将本体图或多本体图中的顶点映射成实数,通过两顶点对应实数间的差值来确定它们对应概念间的相似度。最后,将算法分别作用于GO本体和计算机软件本体,通过实验数据对比说明新算法对特定的应用领域具有较高的效率。

Ontology Learning Algorithm Based on BMRM Iterative Ranking

As a structured model for data storage and representation, ontology has become a hot research field of information retrieval, and is used in biomedical, geography, social science and many other fields. In this paper, we raise new ontology similarity calculation and ontology mapping algorithm based on the BMRM iterative ranking learning method. The optimal parameter vector is built by BMRM iterative, and then obtains the optimal ranking function, which maps each vertex in ontology graph or multi-ontology graph into a real number. By calculating the difference between the real number of two vertices, we determine the similarity between their correspond concepts. Finally, the algorithm act on GO ontology and computer software, respectively. By comparison of the experimental data, we show that the new algorithm has a higher efficiency on the specific field of application.