通过定义不可满足概念间的覆盖关系,发现MUPS和MIPS之间的内在关联,从而引出不可满足概念的R—MUPS;给出并证明不一致本体术语集中至少存在一个不可满足概念的R—MUPS就是该术语集的MIPS;利用这一结论,提出基于有序标签演算的R-MUPS算法,采用深度优先遍历原则合并分支计算R-MUPS,同时缓存覆盖概念集合,加快MIPS的求解,实现本体调试.通过概念扩展树与概念R-MUPS算法的等价性,证明算法的正确性并分析其复杂度.最后,利用自动生成本体、现实本体及其扩建本体的数据进行全面测试.实验结果表明:基于R-MUPS的MIPS求解方法能够高效、准确地完成本体调试任务.
In this paper, an inherent relationship between MUPS and MIPS is obtained by defining covering relations of unsatisfiable concepts and developing the definition of R-MUPS of unsatisfiable concept. Then a proof is given that there exists at least one unsatisfiable concept whose R-MUPS is also a MIPS for every MIPS. Based on this conclusion, an R-MUPS algorithm based on the ordered label calculus is proposed. It applies depth-first traversal in merging branches to calculate R-MUPS, and also caches the concepts of covered at the same time to speed up the solving of MIPS to achieve ontology debugging. Through revealing the equivalence between concept expansion tree and R-MUPS algorithm, the correctness of R-MUPS algorithm is proved, and its complexity is further analyzed. Finally, a comprehensive test is performed using automatically generated ontology test sets, realistic ontology and its expansion ontology. The experimental results show that the solving of MIPS based on R-MUPS algorithm can efficiently and accurately complete the ontology debugging tasks.