Web-based Text Mining For Personalized Retrieval System Design And Implementation,

Along with the popularization of computers and explosive growth of computer networks, more and more electronic information appear, which bring us not only more convenient, but the conflict between vast information and little knowledge. How to pick up sententious information from individuating retrive subsystem based on textual mining to fulfill specific requirements? Data mining or knowledge discovery in database (KDD) would be one approach. This paper describes the implementation of an entire index system of the synthetically medical information, provides the customer a personalized index service based on the data mining technology.Personalized index, a service based on the knowledge discovery, involves in a range of fields such as data-mining, knowledge-index, computer linguistics, informatics etc. "Personalized index" is different from the general information index in methodology, purpose of index and evaluation method. The individuality of index is implemented by the hidden, unknown and useful knowledge found in the process of data-mining. It predicts key word combination where users will visit in the future by finding the key word combination pattern inquired by users through mining registered user's visiting log, and implies the users to make choice they want under the guide.Henan people's hospital, the largest comprehensive medical service department of Henan province, has accumulated a lot of clinic experience, research achievements and case study archives these years. The hospital has set up comprehensive medical database to provide conveniently information index service for customers, and provides the customer better index service with WWW information distributed method. The design and implementation of the index system of the comprehensive medical information, was one of the hospital's research projects in 2002.This paper is based on the integrated medical information network project of Henan people's hospital. The paper is divided into six parts. The first part describes the basic technologies which support the implementation of the personalized index platform. It introduces the technologies used by system development such as personalized index, data mining, conjunction rules and web database in details. The second part explains the overall system layout and subsystem selectivity with the need of the project. As the core topic of this paper, it discusses the classical frequent itemset algorithm of personalized index subsystem. The third, fourth and fifth parts introduce the implementation of three subsystems respectively: database background typing administration subsystem, general index subsystem, and registered user personalized index subsystem. The six part focuses on the discovery of conjunction rules and the forecast of personalized applications.