Query Model Of Rough Relational Database And Its Application

The relational database system has been approved and applied in many fields based on the relational database model during the past several decades. However, it also has some disadvantages that it can't deal with uncertain data so well. With the extension of applications and the sharp expansion of data in quantity, a considerable variety of uncertain, incomplete and vague information has appeared. Hence, it is greatly necessary to generalize the definition of the relational database model. The Rough Relational Database model(RRDM) which combines the Rough Sets Theory(RST) and conventional relational databases had put forward in 1993, so that people has make notable progress in dealing with uncertain problems. Many properties of conventional relational databases don't adapt to Rough Relational Database(RRDB) because the non-atomic trait of attributes values in RRDB. So it is necessary to research the RRDM further.Based on the background above, firstly, the research of current situation of the rough relation database is introduced in this paper. Secondly, according to data characteristics, a method for storing uncertainty data has put forward. According to the attribute values in RRDB data can be partitioned into some equivalence classes, and these classes are operating objects in the fields of application of RRDB, so the storage of equivalence classes should be considered. In this paper a technique for storing the equivalence classes by using Adjacency list is presented. At the same time, a basic table consisting of non-atomic value is stored by using Orthogonal list.The string matching method is used popularly in the conventional database query, and it is acceptable in the rough relational database, but searching efficiency is poor. At present, a query method in rough relational is to partition a table in RRDB into some sub-tables composed with certain data. Obviously, it does not take sufficiently the advantages of RRDM, namely, equivalence classes and concepts on the lower and upper approximation of rough sets have not considered. Thus, a new method is put forward in the paper. The notion of proposed method is to calculate the similarity between the data entered by users and the data being queried in the rough relation database based on the concepts of upper and lower approximation of rough sets. In order to improve the query performances, the data of the rough relational is indexed.The indexing technique in conventional relational database has become popular nowadays. But because of the study on RRDB being in elementary phase, some problems on the indexing technique in the rough relational database have not resolved effectively. In this paper, we show the our concern about the indexing techniques, and an approach to indexing uncertain data is put forward based on the improved formula of Hamming distance. The idea of the proposed approach is to calculate the distance between tuples in the RRDB and to construct a distance matrix. On the basis of the distance matrix, the tuples in the RRDB are classified and the indexing of tuples is implemented.Finally, based on the theory methods above, the query model of the rough relational database is constructed; an algorithm for query of the rough relational databases is described and illustrated by using a soil analysis example. The real world example shows that the proposed approach is useful and effective.