Rule Acquisition In Incomplete And Inconsistent Decision-making Information System

With the continuous development of database systems, efficient data query anddata statistical become possible. But in many cases, for a lack of awareness of thepotential value of mining data, we cannot get potential relationship sand rules inmassive data, leading to the phenomenon of "data explosion, at the same time, a lackof knowledge." So knowledge rules mining techniques have come into being. We can"find the hidden knowledge rules from data mines nuggets". We put the pressure datainto knowledge by using knowledge and rule mining techniques. Therefore, researchthe knowledge of rules mining has great practical significance for us.Classical rough set theory is proposed by Polish mathematician Pawlak in1982.The feature of the theory is that it does not require any prior knowledge and additionalinformation; we can only use the raw data to attribute reduction and rule acquisition.So there are good prospects in data mining, knowledge discovery, etc. Classical roughset theory is mainly used in complete information system, which is use of indiscerniblerelation divided the object into the upper and lower approximation. However, in reallife, because of the data measurement errors or data acquisition restrictions and otherreasons, it makes the phenomena of data uncertainty abound. Incompleteness anduncoordinated of the information are two uncertainty different phenomena. Uncertaintyof information is the essential characteristic of intelligent questions, either humanintelligence or artificial intelligence, is inseparable from the treatment of uncertaintyinformation. Use the classical algorithms to process the data, is bound to affect ourdecisions, resulting in distortion of information. Hence, we need to extend the roughset theory and then applied it for handling of the incomplete, inconsistent informationIn short, how to deal with the uncertainty of knowledge becomes an important issue indecision system.The main contents of research are as follows: First, we overviewed of the development of rough set theory, characteristics,applications and basic theory.Summarized and presented three causes of processinguncertainty and incomplete information, and individually analyzed the characteristicsand scope of these three methods. At last, we respectively analyzed the research statusof incomplete and inconsistent information systems, summed up the commonly usedmethods of problem of handling missing data.Secondly, we detailed research and analysis the classic attributes reductionalgorithm details, and then we introduced the method of information processing inincomplete information systems and uncoordinated information systems, at last weanalyzed and compared the algorithms.Then, classical rough set is based on available information integrity; it ignores thesituation that the information may miss. In this chapter, we give a classic attributereduction algorithm in incomplete information system: Based on the positiverelationship between the domain and heuristic algorithms, and briefly introduced theadvantages and disadvantages of the algorithm.In reality, a lot of the decision-makingtable is dynamic, the new data will be added regularly any time. Incremental rulesmining algorithm updated dynamically, it meets the rules strategies adjust anytime, andit keeps the consistency of the rules in a dynamic environment. Based on the analysisof the advantages and disadvantages of the algorithm, we proposed a new attributereduction algorithm, firstly,we establish the distribution matrix that can identify fromthe original decision table, then we acquire all the distribution reduction set, thefunction is generated by the resolution conjunctive term, derived corresponding ruleand examples of validation. The advantage of this algorithm is it introduction of theconcept of incremental learning, avoids the tedious repetitive calculations. Finally, wecarry out experiment about the positive domain reduction algorithms and heuristicsunder incomplete information system, for the result we verify the feasibility of thealgorithm.Then, we compare and analysis the advantages and disadvantages of the threeclassical reduction algorithms in inconsistent information system. On the basis of theabove algorithm, we use the maximum distribution reduction algorithm to attributereduction in inconsistent decision information system. Then, through thedecision-making and decision-making matrix discernible matrix functions to extract decision rules, we can mining the implicit rule that has credibility in the inconsistentdecision information systems. Through examples illustrate the effectiveness of thealgorithm, to some extent it makes up for the lack of knowledge in informationsystems deficiencies. By comparison experiment with other algorithms to test thefeasibility of the proposed algorithm.Finally, we outlined the work in this article. For the lack of the content of thispaper, we proposed future research directions...