亲属关系推理系统研究与实现

Research on and Implementation of the Kinship Inference System

从亲属关系知识表示和亲属关系推理规则两个方面看,现有亲属关系推理系统缺乏对亲属关系及其形式结构方面的理论探讨。亲属关系推理具有复杂性:亲属关系推理的实质是血缘关系的推理;血缘关系结构是一个以"父亲"、"母亲"、"孩子"组成基础单元的递归的图结构;亲属关系的定义是按照血缘关系结构图中最短路径来给出的;利用婚姻关系、生育关系、被生育关系、兄弟姐妹关系作为四个基本血缘关系,重新设计了亲属关系表达式;利用亲属关系式能够确定系统边界,以及形式化描述亲属关系的相关概念。在此基础上,基于血缘关系结构图的亲属关系推理系统得以构造,利用定位、匹配、检索最短路径和移位四项操作,来完成复杂的亲属关系推理任务。跟以往的亲属关系推理系统相比,此推理系统的逻辑和步骤更加直观简明,保证了亲属关系推理的完备性与准确性。基于血缘关系的亲属推理算法能推广到其他语言的亲属关系推理。

Seen from the point of view of the two elements of knowledge representation and kinship inference rules,the existing system of kinship inference lacks theoretical discussion of kinship and its forms and structures.Kinship inference-essentially the inference of blood relationships-is complex.The structure of such relationships is a recursive graph in which the basic unit,consisting of "father," "mother," and "child," is defined as the shortest path in the mapping of consanguineous structures.The expression of kinship is redesigned using marital,filial,parental,and sibling relationships as the four basic blood ties,with kin relationships being used to determine system boundaries and formally describe the relevant kinship concepts.On this basis,we constructed a kinship inference system based on the mapping of consanguineous structures,carrying out the complex incomplete task of kinship reasoning by using the four operations of positioning,matching,retrieving the shortest path and transposition.Compared with previous kinship reasoning,the logic and procedures of our new system are more intuitive and concise,thus ensuring the completeness and accuracy of kinship inference.Relative inference algorithms based on blood ties can be extended to kinship inference in other languages.