Dublin Core in a Box

Mikhail E. Shvartsman, Alexey S. Ilyin
Russian State Library

1. Introduction

Due to the rapid growth of the amount of Internet resources, librarians face the task to assist readers with finding their way through the new information space. For this they need to understand what can be found in the Internet and which part of it is needed by the library readers. Despite the great number of search engines and quick references, librarians still continue to maintain their own catalogues of selected resources.

This is for the following reasons:

• it is more difficult to find your way through a large catalogue
• A large catalogue already has a system of classification, and it has to be used, though the librarian might not find it satisfactory.
• Large catalogues do not contain resource information needed by the librarians
• Large catalogues usually do not have the things that are necessary for the readers.

Though in many systems there is a possibility of additional input of the Internet resource descriptions, only the web masters, and not librarians, use it to increase the number of visits to their sites. From the psychological point of view it is understandable: it is difficult to put a lot of effort into the updating of "АУ" or "RAMBLER". Nobody knows what will happen to these tomorrow, all efforts may be wasted.

Therefore librarians create small catalogues themselves. Some as HTML - files on their home page, some as a great number of bookmarks in the browser. When the number of resources was small this method worked, however with the increase of the number of retrieved resources there is the growing need to optimise the presentation of search results. Earlier bibliographic services of libraries kept card files of the fulfilled requests, now the time has come to organise a similar system at a new level.

In 1998 the Russian State Library (RSL) with the support of the Russian Foundation for Basic Research (RFBR) started the development of a software application package (SAP) for the support of a systematic catalogue of Internet resources.

In the beginning we defined the main requirements of the application package

2. Application package installation

The application package requires the installation of a computer with Windows 95/98 or Windows NT4.0 operating system. The RAM should be no less than 32 Mb, the application package takes up 30 Mb of a hard disk. The application package includes an installation program, which sequentially will take the user through all stages of the installation. The user only needs to select the type of installation (complete, partial, sample), the directory for storage and the URL or IP address of the computer. To carry out information input and search through the Web the computer should have Internet connection and WEB server software installed. The package includes free software Personal Web Server for Windows 95/98 and a number of free database administration modules.

3. Resource description format

The format Dublin Core Metadata Set was chosen as the format of description. The choice was not very wide. The options were: to use a variety of MARC format, to create a format of our own or to use the Dublin Core Metadata Set.

At present there are contradictory opinions about the Dublin Core Metadata Set. Some consider it a universal format and suggest using it for cataloguing all electronic publications. Others point out the limited number of fields in this format and say that it's impossible to use it.

The truth in our view is as always in the middle. For each format it is necessary to define the area of usage. The same way as there is no universal cure there is no universal format. Limitation of the set of fields leads to the simplicity of descriptions. We think that a resource description format should be simple enough to be understood by any person without special training. From this point of view the Dublin Core Metadata Set is the optimal choice. The implementation of the Dublin Core in Russia is only starting and more detailed information about it can be found on a special Web-site: http://www.rsl.ru/dc.

In the Russian State Library our application package is used for cataloguing of resources by the creators of those resources. For this purpose we have developed an interactive form with a set of fields according to the Dublin Core Metadata Set format. Any creator of a resource can fill in the fields of the resource description and send it to our database. In exchange they receive the description of their resources in Dublin Core Metadata Set tags. They only need to cut the description and insert it into their resource between <head> and </head> tags.

Picture 1. Short form of Internet resource description.

Initially the idea of our project was not only to create an application package for resource cataloguing, but also to make a tool for the generation of metadata in the Dublin Core Metadata Set format. Therefore resource descriptions can be displayed as a table with descriptions.

or as metadata in the Dublin Core Metadata Set format

<head>

<!-- DC RSL, © 1998,2000 Russian State Library -->

<META NAME="DC.Title" CONTENT="О решениях сингулярно возмущенных задач, имеющих пограничный слой типа всплеска. "Фундаментальная и прикладная математика". Выпуск 1, 1995 г.">

<META NAME="DC.Creator.PersonalName" CONTENT="А.Б. Васильева">

<META NAME="DC.Subject" CONTENT="22.151.3">

<META NAME="DC.Description" CONTENT="Рассматривается сингулярно возмущенное уравнение второго порядка. Задается краевое условие, при котором решение имеет пограничный слой типа всплеска. Исследуется асимптотика такого решения по параметру возмущения и его устойчивость.

Постскрипт статьи (без рисунков) (51Kb)">

<META NAME="DC.Publisher.CorporateName" CONTENT="МГУ. Механико-математический факультет">

<META NAME="DC.Publisher.CorporateName.Address" CONTENT="www.math.msu.su/koi8/welcome.htm">

<META NAME="DC.Contributor.PersonalName" CONTENT="Пентус Анна">

<META NAME="DC.Contributor.PersonalName.Address" CONTENT="apentus@mech.math.msu.su">

<META NAME="DC.Contributor.CorporateName.Address" CONTENT="fpm@mech.math.msu.su">

<META NAME="DC.Date.Creation/Modification_of_present_form" CONTENT="21.06.97">

<META NAME="DC.Type" CONTENT="Text, Serial, Journal">

<META NAME="DC.Identifier" SCHEME="URL" CONTENT="http://www.math.msu.su/fpm/koi/95/951/95105.htm">

<META NAME="DC.Relation.Creative" CONTENT="ФУНДАМЕНТАЛЬНАЯ И ПРИКЛАДНАЯ МАТЕМАТИКА 1995, ТОМ 1, ВЫПУСК 1, СТР. 109-122">

<META NAME="DC.Relation.Inclusion" CONTENT="www.math.msu.su/fpm/koi/95/95_1.htm">

</head>

4. Search by hierarchical classification.

For the organisation of search in any hierarchical classification system the application package has a special tool for loading of subject headings from a text file. We have developed a very simple data structure, which is applicable to any hierarchical classification.

The user creates a text file with the list of headings specifying for each heading its code, title and code of a higher level heading. Then this file is processed by a special program, which draws up a tree in accordance with the given relations. This tree is used for navigation similarly to the tree of directories in the Windows 95/98 Explorer.

Having loaded data for the classification system the user can enter codes of headings into the fields of the resource bibliographic description. After the input it will be indicated in the classification tree how many records are contained under headings corresponding to each branch of the tree.

As an example of this facility below there is a text file of the necessary structure with BBK headings for public libraries.

Picture 2. Fragment of BBK for public libraries, numbers of records for each subject are given in brackets.

6. Software structure.

The application package exists in two versions:

• Microsoft SQL Server 7.0 based database management system
• DELPHI database management system, based on Borland Database Engine and PARADOX tables.

The software is distributed free. All the latest modifications will be published at www.rsl.ru/dc. It is necessary to bear in mind that for the use of the Microsoft SQL Server 7.0 version that package should already be installed. The DELPHI version already includes all the necessary components.

Picture 3. Scheme of software operation.

The DC RSL package realises the subset of the Dublin Core Metadata Element Set specification and allows the cataloguing of Internet resources with automatic generation of metadata according to this specification.

The package includes the following components:

• a set of CGI-programs operating on the server side and providing for input, search and mapping of database information;
• service utilities for review, check and maintenance of database integrity.

The software was developed in the Russian State Library with the use of Inprise Delphi 5.0.

PdxRbld.exe program was developed by Roman Crejci, Czech Republic http://www.rksolution.cz

DbExpl32.exe program was developed by Fedor Kozhevnikov, Igor Pavlyuk and Sergei Kovalev from the "Master-Bank", Moscow http://rx.demo.ru

All the other programs use freeware-components of external developers available at Torry Delphi Pages http://www.torry.ru

The programs in the package have the following functions:

CgiDcmes.cgi

CGI-program for the database search by keyword (key /f in URL). It is activated in the search form via f_srch*.htm

DcForm.cgi

CGI-program for information input in the database. The input form is generated off hand with the use of f_*.htt templates containing macrocommands for the entry of database field values.

DcBbk.cgi

CGI-program for database search by BBK indices. Pages are generated off hand with the use of bbk*.htt templates.

PdxRbld.exe

A utility for checking and correction of Paradox tables physical structure.

DataChk.exe

A utility for checking and correction of logical integrity of the database.

Dсrslmgr.exe

• Interactive database editor – allows browsing and deletion of unnecessary data, displaying statistics for the number of records input by operators.
• Interactive BBK index editor – allows export/import of data to/from a text file.
• Program for the search for duplicated records in the database.

7. Further plans for software development

We are working on improvement of our package and the users will be regularly provided with updates.

A lot of libraries in the world already catalogue Internet resources using the Dublin Core format as an internal data presentation format. A number of libraries have joined their efforts to create distributed catalogues. For a simultaneous search in all catalogues of the distributed system the Z39.50 protocol is used. With regards to this a decision was taken to create a freely distributed Z39.50 server for the organisation of distributed databases with the capability to interact with he foreign catalogues.

To achieve this we plan to complete the software development by the end of year. Then it will be transferred to anybody who wishes to have it. Any organisation that wants to maintain a catalogue of Internet resources can receive our software. If the software is installed on a computer with Internet connection the user will receive the database for cataloguing; the interactive forms for cataloguing and search will be published on the user's Internet server. We will also provide a list of addresses of all databases operating with a similar format and the client-program for simultaneous search in all bases. The address of the user database (if necessary) will be communicated to all users of similar databases.

The approach described above is already implemented in some countries. Among the most successful are digital library projects in Bielefeld and Goettingen (Germany), and the European projects MATH NET and METAPHYS.

If we successfully complete our work and assertain the support of the Russian Internet community we can become a part of the global system of distributed catalogues, which will , undoubtedly, promote the development of science.