摘要
人类的认知中具有粒化特性,并且同一现象在不同粒度上具有不同的解释.流图为知识的一种表示形式,素有直观性、计算便捷性和并行处理等特征.以属性.值形式的信息系统作为研究对象,针对新属性的添加而诱导的粒度变化,研究流图在不同粒度上的具体演变.流图在新粒度上的有效性取决于所涉及的等价类的变化和Markov性质的成立.具体的,若新粒度上仅有部分等价类中的成员保持Markov性质成立,则粒度变化可将图形结构由一个粒度上的流图转化为新粒度上的用于构成完整流图的基本构件;若Markov性质在新粒度上不成立,则流图可被转化为新粒度上的与流图无关的结构;若新粒度上等价类中的每个成员皆满足Markov性质,则流图在新粒度上保持不变.流感病人信息系统在不同粒度上的具体分析进一步验证了理论结果.这些结论有助于理解和刻画知识与粒度之间的关系,为模拟人类学习和思维奠定基础。
Granulation is an inherent property of human cognition and the same phenomenon has different interpretations at different granularities. Flow graph is treated as a form of knowledge representation, and is known for its intuitive formation, straightforward computation and parallel processing. Taking the attribute - value information system as the research object, this paper studies the specific changes of the flow graph at different granularities which are induced by adding the new attribute(s). The validity of the flow graph at a new granularity depends on the change of the equivalence classes involved and the establishment of the Markov property. Specifically,if only parts of elements of the equivalence class at the new granularity maintain the Markov property, the change in granularity will then cause the graphical structure to be transformed from a flow graph at one granularity to a basic component for a complete flow graph at this new granularity. If the Markov property does not hold at the new granularity,the flow graph will be transformed into a structure that is unrelated to flow graph at this new granularity. If every element of the equivalence class at the new granularity satisfies the Markov property, the flow graph at one granularity will then remain unchanged at this new granularity. The illustrations of an information system on patients suffering from flu at different granularities further validate the proposed theoretical results. These conclusions can help to understand and characterize the relationship between knowledge and granularity,and lay the foundation for simulating human learning and thinking.
作者
姚宁
苗夺谦
张远健
康向平
Yao Ning;Miao Duoqian;Zhang Yuanjian;Kang Xiangping(Department of Computer Science and Technology,Tongji University,Shanghai,201804,China;Key Laboratory of Embedded System & Service Computing,Ministry of Education of China,Tongji University,Shanghai,201804,China)
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2019年第4期519-528,共10页
Journal of Nanjing University（Natural Science）
基金
国家重点研发计划(213)
国家自然科学基金(61673301,61573255,61573259,61673299,61603278)
公安部重大专项(20170004)
关键词
流图
Markov性质
等价类
粒度
粗糙集
flow graph
Markov property
equivalence class
granularity
rough set
