??????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
In the information technology development today, library, especially the university libraries should not only simple digital conversion and management of information, but also to the archives management and archiving of emerging thing of network, including document translation, text, picture information, audio-visual, multimedia teleconferencing. So the Internet Archive Management, become inevitable trend of the library management, it must be on the problem of archives management of technical and legal related in-depth exposition and discussion.
The so-called data mining (Data Mining), is from the large, incomplete, noise, fuzzy and random data, in which the extraction of implicit, previously unknown but potentially useful information and knowledge. These data may be structured, such as data in a relational database, and can also be semi structured, such as text, graphics, image data, and even heterogeneous data distribution in the network. The method can be found knowledge of mathematics, can also be non mathematics; can be interpreted, can also be summed up. Discovered knowledge can be used for information management, query optimization, decision support, process control, maintenance can also carry on the data of its own. Data mining research achievements over the years by means of mathematical statistics technology and artificial intelligence, knowledge engineering and other areas of the construction of the theoretical system of its own, is a cross subject involving database, artificial intelligence, statistics, mechanics, artificial neural network, visualization, parallel computing etc, is one of the most advanced research directions of the current international database and decision support field.
?????????
A, data mining function
?????????????????????????????????????????????????????????????????????
Data mining by predicting the future trend and behavior, knowledge based decision making predictive. The goal of data mining is to find hidden, meaningful knowledge from the database, its functions can be divided into the following categories.
1?????
1, correlation analysis
??????????????????????????????????????????????????????????????????????
Correlation analysis to find related data in the database, a technique commonly used for association rules and sequential patterns. Association rules are discovered one thing and the other things the mutual relevance or interdependence.
2???
2, clustering
?????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
The input data does not have any type of marker, cluster is according to certain rules to divide the data for the reasonable collection, about to group objects into classes or clusters, so that has a high similarity between objects in the same cluster, and objects in different clusters. Clustering increases people's understanding of the objective reality, is a concept description and deviation analysis prerequisites. Clustering techniques mainly include pattern recognition method and mathematical taxonomy of traditional.
3??????????
3, automatic forecasting trend and behavior
??????????????????????????????????????????????????????????????????????????????????????????
Data mining, automatic classification and prediction in a large database, finding predictive information, automatically is presented to describe the important data model or predict the trend of future data, so before the need for a large number of manual analysis problem can now quickly and directly by the data the conclusion.
4?????
4, concept description
????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
For the complex data in the database, is expected to form a concise description to describe the pooled data set. The concept is to describe certain types of objects are described and summarized the connotation about the characteristics of this kind of object. Concept Description Description Description for the characteristics and difference, the former describes the common features of a class of objects, the difference between the latter describe similar objects. Characteristic generates a class only relates to all objects of the class object in common. Many methods to generate the difference described, such as genetic decision tree method, algorithm.
5, deviation detection
????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
The data in the database often has some abnormal records from the database, the detection of these deviations is very meaningful. Deviation includes many potential knowledge, such as abnormal instance classification rules, does not meet the special case, observations, and the deviation of model prediction value changes with time etc.. The basic method is to look for the observed deviation detection result and the reference meaningful difference value between. This is often used to detect financial fraud in the banking industry, analyses the special consumer spending habits or market analysis.
?????????????????????
Two, the application of data mining in the modernization construction of university archives.
1???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
1, the resource data includes the collection file after digital processing and produce various kinds of electronic archives, electronic records center in all kinds of electronic archives, archival storage software collected information, archives information network construction and maintenance information. We start from the research of university archives user's information requirement, data mining provides a method for University Archives mastering and accurate understanding of Archives Users Information demand.
(1) ??Web???????????????????????Web????????????????????????????????????????????????????????????????????????????????????????????????
(1) using Web to access information mining techniques to discover the connection mode, series mode and Web access trends, user interest model constructing multidimensional view. To determine the file information or service popularity, the discovery of user access patterns and user demand trend, to research the users from different side information needs, providing scientific basis for the construction of the information resources of archives optimization of archives.
(2) ???????web?????????????????????????????????????????????????????????????????????????????????????????????
(2) the collection of University Archives Network Web server reserved user registration information, access to records, and information about users interaction with the system as the original data, after washing, concentration and conversion to form statistical analysis of user access to databases, log database, user customized information database, the user feedback information and other data collection.
2???????????????????????????????????????????????
Starting from 2, the information resources construction of university archives, data mining provides an important basis for choosing a scientific development road of University archives.
(1) ??????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
(1) the mining archives network and file management software to access the information analysis of the utilization of archives resources rate, will use the traditional carrier archives rate is high, high demand priority digitization. For example: Based on the analysis of archival information access records, the search request user request failure data, according to the class of statistical archives rejected for sets and frequent use of set, combined with the aggregation algorithm found collection resource gaps, targeted to supplement and enrich the information resources of archives.
(2) ????????????????????????????????????????????????????????????????????????????????????????
(2) in the process of university archives management by using text mining, association, classification, clustering and other methods, from the massive archives information in accordance with the relevant special subject to mining, classification, processing, finishing and orderly restructuring, the construction of archival information database and all kinds of special archives information library.
3????????????????????????????????????????????????
Starting from 3, do a good job of management of university archives information perspective, data mining for the optimization of collection information and prediction of future work play an important role.
(1) ?????????????????????????????????????????????????????????????
(1) in using link, the user each time borrowing information for association analysis, association rule discovery or proportion of all kinds of archives information relationship, this can further optimize the collection information.
(2) ?????????????????????????????????????????????????????????????????????????????????????????????????????????
(2) the development of university archives information text feature establishment, feature extraction, feature matching, feature set reduction and model evaluation work, the realization of the large collection of documents of the content of the summary, classification, clustering, association analysis, distribution analysis, through the induction and conclusion, the discovered knowledge can be predicted for the trend of archives work in the future.
????????????????
Three, the application of data mining in management in data
???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
Management data of University Archives include: intelligent monitoring system, fire protection system, temperature and humidity control system, intelligent, data management system, data management system etc. by generating a large amount of data in daily work. We need tools to extract valuable knowledge in this kind of useless data and its application to the university archives work in data mining, and play a role in the modernization construction of University archives.
The focus of university archives work is the service for teachers and students, to serve as the center to carry out the work, how to use the advanced tools, improve the quality of services is baffled us. The data provide effective mining method for the university archives work intelligent, personalized, quality. In intelligent retrieval system can call the user interest model, automatic correction of retrieval strategy and based on user interest will be the retrieval results fast clustering and classification, and organized to sort out; for Design Institute, Academy of Social Sciences and other scientific research archives users, can use data mining to carry out targeted archives information mining, and apply the research results to outline, results report form is provided to the user. It not only realizes two time development of university archives, also will give the user to bring unexpected surprises.
Exchange of documents between the network initially scientists and researchers of the software, the Internet for education and research can get government subsidies. In Chinese, university funding, with university support library, digital library network archives is not profitable, output is the long-term social benefits of teaching and scientific research. Today, the Internet has become more and more commercialized, the network has become a potential technology investment object in the digital economy. Digital Library of university can also consider establishing for-profit network archive, using some business mode of the network business, such as network advertisement, banner ads, sponsorship advertising, subscriptions, B2C etc.. Income can be used for rolling development of university network construction of Digital Library's archives. At present people on these budding economic model know little. Public policy making body is the network management of government departments, the implementation of e-government, the development of cyber source, promote the transfer of publishing from the text printing to the network is an important task for the relevant government departments. Policy, attitude and measures of university is crucial to the development of digital library. The market means and policy of balance is the museum construction of network archives, archives operation of network, online content delivery and preservation should and must be considered.
?????
Reference.
[1]Peer to peer Networking and Digital Right Management?by Michael A. Einhorn, Bill Rosenblatt, Policy Analysis No.534, CATO Institute. Fabruary 17,2005
[1]Peer to peer Networking and Digital Right Management, by Michael A. Einhorn, Bill Rosenblatt, Policy Analysis No.534, CATO Institute. Fabruary 172005
[2]What Every Citizen Should Know About DRM, aka Digital Right Management, by Mike Godwin Seuitoo Technology Counsel At Public Knowledge, 30 January, 2008, Ebook-Computer & Internet,
[2]What Every Citizen Should Know About DRM, aka Digital Right Management, by Mike Godwin Seuitoo Technology Counsel At Public Knowledge, 30 January, 2008, Ebook-Computer & Internet,
[3]??·??.???????.??????2004?13?
[3] Peter Lyman. Think of the World Wide Web archives. Information reference, 2004 (13)
[4]????.??????????14????.??????2008?3?
[4] Zhou Hongren. 14 key points of information to help build a harmonious society. China information industry, 2008 (3)
[5]ITU-T Technology Watch Reports. 2006-2008, Telecommunication Standardization Policy Division, ITU Telecommunication Standardization Sector
[5]ITU-T Technology Watch Reports. 2006-2008, Telecommunication Standardization Policy Division, ITU Telecommunication Standardization Sector