Rough Set Model Based On Weak Similarity Relation In Incomplete Interval-valued Information System

With the development of information technology,the amount of data obtained in various fields is huge and the form is chaotic.New requirements for data analysis are put forward.How to extract and extract effective knowledge from big data is one of the hot topics at present.Rough set theory which can effectively depict the uncertainty of knowledge,is a kind of deal with imprecise,inconsistent,mathematical tools for incomplete information,due to the rough set theory based mathematical maturity,does not require prior knowledge,but also have very strong complementarity with other uncertain problems in data analysis and knowledge discovery applications.The interval value can present the uncertain data effectively in real life,for instance,a period of temperature and water quality monitoring results often appears in the form of interval value.However,the incomplete data are often inevitable which may be caused by the loss of information,omission,measurement error,data noise,the fault of transmission medium and others.Because the incomplete interval-valued information system have characteristics both incomplete and imprecise,the extended rough set model for incomplete interval valued information will help to fully understand data and discover knowledge.In real life,due to different needs and basic conditions,the selection of models and data processing are not the same,both the objective and subjective need to be considered.Therefore,dynamic knowledge discovery and data analysis are particularly important.In this paper,a new model is constructed for incomplete interval valued information systems based on rough set theory.(1)From the perspective of collection and probability,after studying binary relation in complete interval valued information systems,we build compatible similarity relation and achieve compatibility between complete and incomplete in interval valued information system.Based on the tolerance relation,an extended rough set model is constructed,then the rationality of the uncertainty measure is verified by the experiment.(2)From the distance perspective,through the analysis of incomplete interval valued information systems,we studied the similarity relations from different angles,then propose the concept of weak similarity relation and prove the rationality of weak similarity.Based on the weak similarity relation,an extended rough set model is constructed,and the rationality of the uncertainty measure is verified by the experiment.(3)In the rough set model in incomplete interval valued information system based on weak similarity,after analyzing the information system,we propose an adaptive learning algorithm of tolerance degree.The algorithm makes full use of the hidden information system the knowledge itself,which providing another way for tolerance degree in the rough set model.