Study And Implementation Of A Data Cleansing System Based On Multi-Agent Technology

Currently, the extent of enterprise information is increasingly demanding high-level, in which an important aspect is the management of enterprise data. Based on the principle of "Garbage in, garbage out", the data which is managed must be reliable, no mistake and truly reflecting actual enterprise situation for supporting to make right decision. Therefore data cleansing acquires increasing attention, because data cleansing is a significant method to improve data quality. However, current data cleansing systems still face with many problems, agent technology is used to solve the problems faced by effective way.The agent technology originates in Artificial Intelligence, an agent is an autonomous entity, and it is situated in some environment, is capable of flexible, autonomous action in that environment in order to meet its design objectives, and can also influence that environment. Generally speaking, the functions of an agent are limited. Hence, applications of multi-agent system are more popular. In a multi-agent system, information about the environment should be exchanged between agents, through consultation and cooperation to realize the function of the system.This thesis firstly analyzes the status quo of data cleansing and gives the problems of current data cleansing systems:poor intelligence, lack of interaction, lack of data analysis functions, limited capacity to deal with the distributed data cleansing and the real-time data cleansing. Secondly, this thesis introduces the definition of agent and some of its major characteristics from the agent's basic theory, and then gives a feasibility demonstration of using agent technology to solve the above problems. On this basis, this thesis proposes a data cleansing model and a data cleansing framework based on multi-agent, through the detailed description of the system's implementation demonstrates how to use agent technology in the field of data cleansing. Using the ideas and methods of agent technology in data cleansing, it can increase the intelligence of data cleansing system, expand the system's interactive capabilities, reduce the user's repetitive work, and enhance the efficiency of the data cleansing.Finally, data cleansing system based on multi-agent is applied on ocean data environment. Through specific application, this thesis verifies that Data cleansing system has good detection and control capability at the accuracy, effectiveness, integrity and consistency of data, and verifies the practicability of data cleansing system based on multi-agent.