Computer Science 代写 Fuzzy Object Relational Approach
PROBLEM STATEMENT:
In this kind of business searching process a set of characteristics is specified for the real estate and banking to have, but usually these characteristics are not fully defined. A customer has a set of favorites, a general idea of what is being looked for, that idea not necessarily should fit to a crisp value, it might be most accurately represented by a value range, an approximate value or even an upper or lower bound. The imprecise representation of these characteristics may allow obtaining results that verify our preferences on different degrees.
PROBLEM DESCRIPTION:
Generally the fuzziness is managed by sales agents who can easily process and handle fuzzy the sale agent, both of them can handle fuzzy information naturally.
The problem arises when one of these entities, the sales agent in our case, capable of handling fuzzy information, is replaced by an automatic system. It is necessary to provide the system with methods to handle fuzzy information in the same way the sales agent was doing before. So, a way to represent fuzzy information about real estates is proposed, in order to be able to design a system that can mimic the sales agent behavior, to interact fluidly with a customer.
The proposed case here is to represent fuzzy information in an object-oriented data model, and recent work point to a model and implementation of a FORDBMS using the object features of current ORDBMSs to extend them by means of UDTs, which encapsulate fuzzy information representation and processing.
FUZZY OBJECT-RELATIONAL DATABASE
An ORDBMS can be extended considering user defined data types (UDTs) to virtually maintain any kind of complex data, similar to multimedia or spatial data. Extending an ORDBMS with fuzzy data management UDTs, produces a FORDBMS which combines the power of fuzzy groups, and the object oriented and relational models.
This extension provides advantages over current FRDBMSs, such as tight level of integration with the fundamental DBMS, hiding implementation aspects of fuzzy types, which allow the user to be aware only of semantics and functionality, an extensible schema allowing future extensions, and efficient implementation, avoiding the use of software wrappers to allow fuzzy data management.
Figure 1 Object-Relational Impedance Mismatch Diagram
The above diagram shows the Object-Relational Impedance Mismatch, People creating object hierarchies are doing something fundamentally different than people creating databases. A similar mismatch applies between people participating in many sites and people consuming them
DATA-TYPE HIERARCHY FOR FUZZY DATA MANAGEMENT
For giving complex fuzzy data management abilities for the fundamental ORDBMS, new UDTs have been defined using host DBMS object-oriented features, organized in a hierarchy and extending the basic DBMS data-types. These new data-types allow the DBMS user to deal with several kinds of imprecise data.
The types in the said hierarchy are the following: -
FUZZY DATA-TYPES (FDT) is concept of all sustained fuzzy data. This type declares common general methods to be applied in the subtypes, for instance the FEQ (fuzzy equal to) method, which extends the concept of classical equality to the fuzzy framework, returning a value in the interval representing the fuzzy resemblance degree between two fuzzy values.
ATOMIC FUZZY TYPES (AFT) collects all common behavior for the fuzzy extensions of scalar and numerical data.
ORDERED AFTS (OAFT) gives structure and behavior to atomic fuzzy data represented by a possibility distribution defined on an ordered domain (numerical fuzzy data). As this type has an associated ordered domain which defines an order relation between the domain elements, the type can define an extension of the classical relational operators, for instance fuzzy equal to (FEQ), fuzzy greater than (FGT), fuzzy greater than or equal to (FGEQ), etcetera.
NON ORDERED AFTS (NOAFT) offers structure and behavior to data defined on a scalar domain without an order relation. The user defines a fuzzy nearness relation between domain's members, which is used to compute the resemblance degree between two members using the FEQ operator.
Fuzzy Collections (FC) extends the classical collection concept to a fuzzy one, in which the collection elements have a membership degree between. Fuzziness affects only elements’ membership, it does not affect collection elements, and consequently gathering elements’ type can be fuzzy or hard. FC type provides the required structure and behavior to manage the collection, like methods for adding, removing or getting the membership of collection elements.
THE REAL ESTATE SEARCHING PROBLEM
Real-estate management process selected as an ideal object example that provides complexity whereas attributes to fuzzy treatment due to the high level of imprecision in their values.
In the real estate searching process a set of characteristics is specified for the real estate to have, but usually these characteristics are not fully defined. A customer has a set of preferences, a general idea of what is being looked for, that idea not necessarily should fit to a crisp value, it might be most accurately represented by a value range, an approximate value or even an upper or lower bound. The imprecise representation of these characteristics may allow obtaining results that verify our preferences on different degrees.
Generally the imprecision is managed by sales agents who can easily process and handle fuzzy information. The real estate management process occurs between two humans, the customer and the sales agent, both of them can handle fuzzy information naturally.
Figure 2 Real-Estate Searching Process [3]
The problem arises when one of these entities, the sales agent in our case, capable of handling fuzzy information, is replaced by an automatic system. It is necessary to provide the system with methods to handle fuzzy information in the same way the sales agent was doing before. So, a way to represent fuzzy information about real estates is proposed, in order to be able to design a system that can mimic the sales agent behavior, to interact fluidly with a customer.
Which attributes are suitable for fuzzy handling will be examined, and also, the way to represent them in the framework defined in previous sections?
