(A) Archive selection and collection This stage determines what information is archived and the extent and depth of storage. Unlike traditional media, there is no fixed rule for the selection of information on electronic carriers, but there are guidelines that can be followed. These guides can help determine the following aspects:
(1) Choose the object of archiving The National Library of Canada and the National Library of Australia have affirmed the importance of the selection guide. Because the content on the Internet is very complicated, it is necessary to determine a selection principle through the guideline. For example, the principle of the Canadian National Library is to select content that has research value and cultural inheritance. The Australian National Library's PANORA project only archives Australian Internet publishing content, but at the same time it also pointed out that it is not possible to file all of this content, so it also provides corresponding guides to choose.
(2) Determination of the scope of the archive Another issue directly related to the selection of the object is to determine the scope. Since not all aspects of a digital work are worth storing, especially for some large and complex Internet sites, full storage may be completely unnecessary, so it is more necessary to define the scope.
Still, taking the PANDORA project of the Australian National Library as an example, the need for content is completely determined by whether or not the link at the upper or lower level of a site is selected. PANDORA's approach is to divide the content of the site into several parts, but they are compared with the requirements of the guides respectively, and those that meet the requirements are archived and vice versa. Of course, if the entire site content has value, it will be fully preserved.
(3) Archiving of Links Hyperlinks are widely used in electronic publishing. Therefore, it is necessary to determine whether these links and their contents are filed together with the original. For this issue, each project responds differently.
Links (URL domain names or other identifiers) are saved in most projects, but the links to the content are not saved. In the case of the American College of Medicine, it preserves all links and reference materials nested in the original text, but does not preserve the contents of the links unless they are also part of the original text or are originally intended to be archived. Similarly, the United States Department of Energy’s Office of Science and Technology Information operates this way, keeping links but excluding external links beyond the archived project.
What is slightly different is the National Library of Canada. In addition to the save link, if the link content is on the same server as the archived text, it will be saved. Of the 18 projects, there are only two exceptions. They save both external links and related content. This is quite relevant to the special purpose of these two projects. For example, one of the projects is to keep a complete picture of the Internet, so it All relevant content is included.
(4) Updating archive content While archiving digital information, these information are often updated constantly. Like some long-term sites, the archived information needs to be updated. So how is the updated cycle and scope determined? This requires finding a balance between the content and completeness of the archives, accuracy and cost, as it is clear that the increase in content updates and archiving content will result in an increase in archival costs. The Australian National Library’s approach is to add an option form to each automatic level of “published†items. Options include: opening/closing, weekly, monthly, quarterly, semi-annual, once every nine months, yearly, One update cycle depends on the expected degree of change and the stability of the entire site.
(II) Collection channels For the collection of relevant network information, there are two basic channels: hand selection and machine selection. In the National Library of Australia project, all sites are manually browsed and selected, and staff will track the continuity of these sites before archiving. In contrast, the Swedish National Library, the Royal Library of Sweden, uses automatic machine selection. The Kultruarw project it hosts regularly collects material through an automated device. There is no need for value judgment. These information collection devices will find information on all known sites in Sweden, and of course also include information on Sweden on foreign sites, such as travel information and online foreign translations of Swedish children's products. Although it is automatically collected, the Royal Library still prefers to select periodicals, fixed files, HTML pages, and references, user network groups, ftp archives, and databases will be considered backwards.
The National Library of Finland, the University of Helsinki's EVA project uses similar technology to Sweden, but the EVA project follows the guidance of the wizard when it conducts automatic collection. In order not to overload the server, EVA limits the time interval between two acquisitions at the same site. Although this limitation is for the sake of project benefits, EVA developers still believe that the mechanism is not complete and flexible. They hope that the specific time limit will be more explicit and implemented at each URL site level. But to achieve this flexibility requires the establishment of a data-based application for this purpose and that can be modified by librarians.
(iii) Identification and cataloguing of digital archives Once the objects of digital archiving have been identified, it is necessary to identify and catalog them. Identification tags provide specific clues for finding these objects, and may link to other objects that are related to them. Catalogs that exist in the form of metadata will provide relevant structure, access, and other information.
All archived projects use some form of metadata to describe, manage, and save the archived objects. Some common questions during the project development process are how these metadata are generated, what the generation criteria are, how their application hierarchy and where they are stored.
Of the 18 projects surveyed, most generated full or partial metadata during the cataloging stage. However, at present, people are increasingly realizing that artificially generated metadata is an obstacle in the process of digital archiving. Therefore, there is an increasing trend toward automatic generation mechanisms. . One of the operations that has attempted this operation is the archiving project carried out by the United States Environmental Protection Organization. It directly derives metadata from the digital element level. Another project, DITT, also invests in the automatic metadata generation system.
In the selected 18 projects, various metadata formats were used. Most national libraries use traditional library cataloging standards. Like the PANDORA project of the National Library of Australia, the MARC cataloging was used for electronic files, but new simplified formats were also used for network information. The EVA project uses a format similar to the Dublin Core. It is expected that a simplified format such as Dublin Core will be obtained directly from publishers and applied to metadata, thus eliminating the need for extensive use of library cataloging.
These 18 projects use a wide variety of content standards in the process from data to metadata. The National Library tends to use traditional standards such as AACR2, and there are existing information standards in some areas, such as longitude and latitude standards, that can be easily incorporated into the metadata content standard. But at the same time, we must point out that for long-term storage needs special elements for special elements, especially some non-text data such as images, videos and multimedia.
The level of use of metadata depends on the data type and the expected access requirements. Data groups are usually cataloged at the file or collection level. Articles in e-journals can be catalogued separately, and sometimes it is not necessary to consider metadata at the headlines of articles and journals. In addition, when considering a home page, it is more troublesome because it is necessary to determine which times the metadata is used on the page.
Of the 18 projects surveyed, the metadata files are usually stored separately from the archive files. The library will place these files in the “public access†directory, and publishers may place them in a bibliography or an introduction database. However, there are some exceptions, such as e-journals that tag titles, authors and other information. Such information can exist with documents and can be extracted for cataloging.
In the discussion of digital archiving, there is a general concern that there is a need to be able to convert and use different metadata formats. For this issue, the OAIS (Open Archival Information System) reference model developed by the ISO Advisory Group has attempted to add specific metadata to each archiving object by its type.
For those situations where the digital material is not directly copied into the archive, the material is transferred from one server to another server on the network or from one directory to another. This conversion brings many changes to the URL domain name due to the URL. The domain name indicates the location of the information. When this identification information changes, the source file and the link file will not be persistent.
Although there are potential issues, most archives continue to use URLs to indicate the location of digitized objects. However, some projects have also been changed. For example, the OCLC archive uses PURL. The American Chemical Society uses a digitized object identifier http:// for its journal articles. It also contains the original number in the catalog at the beginning of the publishing process. . The American Astronomical Society (AAS) uses a large number of identification systems. The naming solution replaced the URL, and AAS also used an aeronautical standard identifier called the Bib code, which has been used in this field for 15 years. In the spring of 1999, AAS added the publication reference number and another logo could be added when it was necessary to include links.