基于结构正则化模型的本体算法

Ontology Algorithm Based on Structured Regularization

通过结构正则化模型的优化得到本体函数,函数空间采用再生核希尔伯特空间,数据项用样本的经验误差表示,利用凸函数的范数求和得到结构扰动项,通过迭代计算得到最优解,并通过参数σ的计算来控制步长。将该算法应用于植物学PO本体和仿生机器人本体,验证了本算法对植物学领域的相似度计算和在仿生机器人领域建立本体映射的效率。

The ontology function is obtained via structural regularization model,function space is used as reproducing kernel Hilbert Space,data term is got from empirical error,disturbance term is deduced by the sum of convex function norms,the solution is achieved in terms of iterative calculation,and the step size is determined by the computation of parameterσ.The algorithm is applied to the PO and humanoid robotics ontologies.It shows that the efficiency of new algorithm for calculating the similarity in plant field and establishing the ontology mappings in humanoid robotics application.