source: other-projects/nightly-tasks/diffcol/trunk/gs3-model-collect/DSpace-To-GS/archives/HASH0164.dir/doc.xml@ 28241

<?xml version="1.0" encoding="utf-8" standalone="no"?>
<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">
<Archive>
<Section>
  <Description>
    <Metadata name="gsdldoctype">indexed_doc</Metadata>
    <Metadata name="Language">en</Metadata>
    <Metadata name="Encoding">utf8</Metadata>
    <Metadata name="Author">Bronwyn</Metadata>
    <Metadata name="Title">Distributing Digital Libraries on the Web, CD-ROMs, and Intranets: ...</Metadata>
    <Metadata name="URL">http://research/ak19/gs3-svn-26Aug2013/web/sites/localsite/collect/DSpace-To-GS/tmp/1378708406/3.html</Metadata>
    <Metadata name="UTF8URL">http://research/ak19/gs3-svn-26Aug2013/web/sites/localsite/collect/DSpace-To-GS/tmp/1378708406/3.html</Metadata>
    <Metadata name="gsdlsourcefilename">import/3/3.pdf</Metadata>
    <Metadata name="gsdlconvertedfilename">tmp/1378708406/3.html</Metadata>
    <Metadata name="OrigSource">3.html</Metadata>
    <Metadata name="Source">3.pdf</Metadata>
    <Metadata name="SourceFile">3.pdf</Metadata>
    <Metadata name="Plugin">PDFPlugin</Metadata>
    <Metadata name="FileSize">218837</Metadata>
    <Metadata name="FilenameRoot">3</Metadata>
    <Metadata name="FileFormat">PDF</Metadata>
    <Metadata name="srcicon">_iconpdf_</Metadata>
    <Metadata name="srclink_file">doc.pdf</Metadata>
    <Metadata name="srclinkFile">doc.pdf</Metadata>
    <Metadata name="NumPages">7</Metadata>
    <Metadata name="ex.dc.Contributor">Ian Witten</Metadata>
    <Metadata name="ex.dc.Contributor">Sally Jo Cunningham</Metadata>
    <Metadata name="ex.dc.Contributor">Bill Rogers</Metadata>
    <Metadata name="ex.dc.Contributor">Rodger McNab</Metadata>
    <Metadata name="ex.dc.Contributor">Stefan Boddie</Metadata>
    <Metadata name="ex.dc.Date^accessioned">2005-01-10T02:51:09Z</Metadata>
    <Metadata name="ex.dc.Date^available">2005-01-10T02:51:09Z</Metadata>
    <Metadata name="ex.dc.Date^issued">2005-01-10T02:51:09Z</Metadata>
    <Metadata name="ex.dc.Language^iso">en</Metadata>
    <Metadata name="ex.dc.Title">Distributing Digital Libraries on the Web, CD-ROMs, and Intranets: Same information, same look-and-feel, different media</Metadata>
    <Metadata name="equivlink"></Metadata>
    <Metadata name="Identifier">HASH016449897ef791600b644989</Metadata>
    <Metadata name="lastmodified">1378708203</Metadata>
    <Metadata name="lastmodifieddate">20130909</Metadata>
    <Metadata name="oailastmodified">1378708406</Metadata>
    <Metadata name="oailastmodifieddate">20130909</Metadata>
    <Metadata name="assocfilepath">HASH0164.dir</Metadata>
    <Metadata name="gsdlassocfile">doc.pdf:application/pdf:</Metadata>
  </Description>
  <Content>
&lt;A name=1&gt;&lt;/a&gt;&lt;b&gt;Distributing Digital Libraries on the Web,&lt;/b&gt;&lt;br&gt;
&lt;b&gt;CD-ROMs, and Intranets:&lt;/b&gt;&lt;br&gt;
&lt;b&gt;Same information, same look-and-feel, different media&lt;/b&gt;&lt;br&gt;
Ian Witten, Sally Jo Cunningham, Bill Rogers, Rodger McNab,Stefan Boddie&lt;br&gt;
Department of Computer Science&lt;br&gt;
University of Waikato&lt;br&gt;
Hamilton, New Zealand&lt;br&gt;
&lt;b&gt;Abstract:&lt;/b&gt; The Greenstone system from the New Zealand Digital Library provides a&lt;br&gt;new way of making collections of information available in the same form over&lt;br&gt;the World-Wide Web, on CD-ROM, or on local Intranets.  Exactly the same&lt;br&gt;information is available in each case, and exactly the same interface is used to access it.&lt;br&gt;The New Zealand Digital Library is accessible over the Web and offers a wide variety&lt;br&gt;of information collections.  Sub-collections can be written to a CD-ROM, which can be&lt;br&gt;used on a standalone PC by a single user.  A local Web browser suffices to access the&lt;br&gt;information on the disk just as though the PC were connected to the Internet.&lt;br&gt;Simultaneously, if there is a network connection, the same disk acts as a network server&lt;br&gt;to make exactly the same information available to others who need only use their&lt;br&gt;standard Internet browser software.  This technology has great appeal for many users,&lt;br&gt;particularly those in developing nations where non-local Internet access can be&lt;br&gt;precarious or prohibitively expensive.&lt;br&gt;
&lt;b&gt;1. Introduction&lt;/b&gt;&lt;br&gt;
The emerging digital library movement is a child of the Internet and the World-Wide&lt;br&gt;Web. Spurred on by visions of an “information superhighway,” current digital library&lt;br&gt;projects invariably concentrate on providing access to document collections over the&lt;br&gt;Internet, where documents, users, and catalog may all be distributed widely.  Often the&lt;br&gt;search interface is WWW-based, in contrast to the telnet or phone-in access required by&lt;br&gt;library OPACS and earlier commercial “online” bibliographic databases such as Dialog.&lt;br&gt;Web-based digital libraries have significant advantages over their online predecessors.&lt;br&gt;Users need not obtain and install search software on their own sites. In many areas&lt;br&gt;Internet access incurs minimal charges, or at any rate is significantly cheaper than a&lt;br&gt;direct telephone connection with the retrieval system. Finally, Web browsers provide a&lt;br&gt;simple, standard means of access to a variety of digital library systems.&lt;br&gt;
However, practical experience in digital library development indicates that in many&lt;br&gt;situations, universal access via the Internet is neither possible nor desirable.  A&lt;br&gt;business, for example, might desire a digital library to make its proprietary documents&lt;br&gt;available to its employees, but only if the company’s security could be ensured by&lt;br&gt;restricting access with an intranet.  CD-ROM has been identified as the implementation&lt;br&gt;platform of choice for collections targeted at large portions of the Third World; for&lt;br&gt;many developing countries, particularly in Sub-Saharan Africa, Internet connections are&lt;br&gt;still either non-existent, undependable, or prohibitively expensive to use.  Despite its&lt;br&gt;lowly status, the CD-ROM has many advantages. Relatively durable in the face of harsh&lt;br&gt;environmental conditions, it incurs known, fixed costs for purchase and supporting&lt;br&gt;hardware (White, 1992). It makes information accessible on a tangible medium that is&lt;br&gt;under the user’s control and is not subject to capricious decisions by others. A CD-&lt;br&gt;ROM based digital library carries the further advantage of providing full document&lt;br&gt;contents—a significant drawback to bibliographic systems being that their users in&lt;br&gt;developing countries could locate descriptions of relevant documents, but were then&lt;br&gt;often unable to obtain the documents themselves (El-Hadidy, 1994; Chowdhury,&lt;br&gt;1996). Finally, while a CD-ROM holds a reasonable amount of material in textual form,&lt;br&gt;digital videodisk technology is already available which can store 12 Gb on a single&lt;br&gt;disk—far larger than most extant textual digital libraries.&lt;br&gt;
&lt;hr&gt;
&lt;A name=2&gt;&lt;/a&gt;For this reason the Greenstone digital library software developed by the New Zealand&lt;br&gt;Digital Library project allows a collection developer to create a  digital library that is&lt;br&gt;WWW-based, intranet-based, or available on a standalone or networked CD-ROM.  All&lt;br&gt;platforms support exactly the same interface, and the same search and retrieval&lt;br&gt;methods. This standardization reduces the system learning curve for intranet or CD-&lt;br&gt;ROM users who have previous experience with WWW browsers, and conversely&lt;br&gt;allows those users currently without Internet access to more easily progress to Web&lt;br&gt;searching and browsing when it becomes available to them.&lt;br&gt;
An earlier version of this software has been used in a university-level distance learning&lt;br&gt;course on computer literacy, where selected portions of various WWW sites were&lt;br&gt;stored on CD-ROM for students to surf (Holmes and Rogers, 1997). Here, the primary&lt;br&gt;advantages of avoiding an Internet connection were to smooth out variable page&lt;br&gt;retrieval times, to avoid problems with off-site servers going down or being temporarily&lt;br&gt;unavailable, and to eliminate communication costs.  In secondary or primary school&lt;br&gt;settings, this technique for capturing known portions of the WWW can be used to&lt;br&gt;prevent students wasting lab time exploring sites that irrelevant to the task at hand, or&lt;br&gt;that are inappropriate for their age groups.&lt;br&gt;
The digital library collection described in this paper is comprised of a set of documents&lt;br&gt;provided by the United Nations University, focusing primarily on food and nutrition.&lt;br&gt;The goal of the United Nations University Press is to disseminate knowledge in the&lt;br&gt;field of the global problems of human survival, development and welfare, in order to&lt;br&gt;increase dynamic interaction in the world-wide community of learning and research.&lt;br&gt;By making their documents available in a variety of formats—print, CD-ROM, WWW&lt;br&gt;pages—this research and human development information can be distributed more&lt;br&gt;widely, and in a form appropriate to the conditions required by  information users.&lt;br&gt;
Section 2 describes the software architecture. Multimedia collections are supported, and&lt;br&gt;a single collection may include text, images, audio, and even video clips. Compression&lt;br&gt;technology is used to ensure that the greatest possible volume of information is packed&lt;br&gt;into a limited storage space. The interface software combines easy-to-use browsing&lt;br&gt;with powerful search facilities. As discussed in Section 3, several ways are provided to&lt;br&gt;find information in a collection; a user can conduct keyword searches, access known&lt;br&gt;documents by title, or browse subject “bookshelves”.&lt;br&gt;
&lt;b&gt;2 . System architecture&lt;/b&gt;&lt;br&gt;
A great advantage of the WWW as a means of presenting and using information is that&lt;br&gt;very little direct user interface programming is required.  A system can generate simple&lt;br&gt;text documents in HTML notation, and leave the task of display, printing, screen&lt;br&gt;navigation, and so forth to a Web browser.  As a result, the browser writer takes most&lt;br&gt;of the burden of system dependence away from the application programmer.  The CD&lt;br&gt;version of the Greenstone library follows this structure:  our software takes the form of&lt;br&gt;a WWW server, communicating with an unmodified browser  using IP networking&lt;br&gt;software.  While the primary goal is to have a system running on a stand-alone&lt;br&gt;machine, the use of IP networking does also mean that the software will function as a&lt;br&gt;WWW server over an external network.  Figure 1 shows the general software&lt;br&gt;organization.  The gray box encloses the software components running on one&lt;br&gt;machine.&lt;br&gt;
Ideally, the WWW server would be a standard piece of software, and a digital library&lt;br&gt;would take exactly the same form on a single machine as it does on our larger WWW&lt;br&gt;serving equipment.  This did not prove possible for a number of reasons—most&lt;br&gt;significant of which was the amount of memory expected to be available on our target&lt;br&gt;machines, which for this project include the older and smaller workstations commonly&lt;br&gt;
&lt;hr&gt;
&lt;A name=3&gt;&lt;/a&gt;in use in the Third World.  The full digital library system on our WWW servers does&lt;br&gt;make use of standard Internet server software.  In the WWW version of our digital&lt;br&gt;library architecture, pre and post processing of queries on the library are handled in&lt;br&gt;tasks run via the CGI mechanism, and communicate via request queues with tasks&lt;br&gt;running the MG document indexing and compression software (Witten et al, 1994).&lt;br&gt;Much of the ‘glue’ software is written in Perl (Wall et al, 1996) and requires the large&lt;br&gt;Perl interpreter and software library to be in memory.&lt;br&gt;
In contrast, the CD-ROM version of the software is a single integrated piece of software&lt;br&gt;incorporating the Web server, digital library pre/post processing,  and MG.  Only a&lt;br&gt;single index need be in memory at any one time, as a CD-ROM usually only holds a&lt;br&gt;single collection.  All of the software is coded in C and C++ to avoid the significant&lt;br&gt;overhead involved in using a Perl interpreter.  The result is a system which will work&lt;br&gt;satisfactorily on a workstation with 8 or 16 MB of main memory (depending on the&lt;br&gt;memory requirements of the workstation’s operating system).&lt;br&gt;
A browser is directed to access the server in one of two ways.  The simplest is to use&lt;br&gt;the URL http://127.0.0.1 (127.0.0.1 means ‘ local machine’).  Once the first page is&lt;br&gt;loaded, further pages are referenced relative to the starting page, and so are also&lt;br&gt;obtained from the server.  This is convenient in that it requires no set-up on the&lt;br&gt;browser.  The alternative is to set the browser to use 127.0.0.1 as its ‘proxy’.  This&lt;br&gt;means that all page requests are routed to the server.  It functions like a fixed cache,&lt;br&gt;satisfying requests when it can and passing demands that it cannot handle on to an&lt;br&gt;external network (if available).&lt;br&gt;
external network&lt;br&gt;
internal network software&lt;br&gt;
BROWSER&lt;br&gt;
 SERVER&lt;br&gt;
Local File Retrieve&lt;br&gt;Local Text Database (MG)&lt;br&gt;Local Non-Text Repository&lt;br&gt;Remote (WWW) access&lt;br&gt;
CD&lt;br&gt;
Special Processing&lt;br&gt;
&lt;b&gt;Figure 1: Browser-Server Interface&lt;/b&gt;&lt;br&gt;
The server handles incoming page/file retrieval requests according to the requested&lt;br&gt;item’s availability and form of storage.  If a page is not available locally, the request&lt;br&gt;may be passed on to an external network. If  each page or document in a collection is&lt;br&gt;stored in a separate file, then a local file request can access the item on the CD-ROM.&lt;br&gt;However, in general we avoid storing a collection’s documents in separate files,&lt;br&gt;because large numbers of files use CD-ROM space inefficiently.   Instead, document&lt;br&gt;files containing text are stored (and the extracted text is indexed) in an MG database,&lt;br&gt;and non-text files are stored in a special repository file. The server has an index of the&lt;br&gt;documents held in the MG database and the file repository.  Incoming requests are&lt;br&gt;checked against this index and may be retrieved from MG or the repository as&lt;br&gt;appropriate. Major savings in collection storage requirements are possible by taking&lt;br&gt;advantage of MG for text storage: typically text compresses to 25% of its original size,&lt;br&gt;and the compressed index occupies around 7% of the size of the original text. This&lt;br&gt;leads to a total storage requirement for the indexed collection of approximately one-third&lt;br&gt;of the size of the original text alone. The system can also support a variety  of types of&lt;br&gt;
&lt;hr&gt;
&lt;A name=4&gt;&lt;/a&gt;non-text items in the collection—audio, images, video clips—simply by including&lt;br&gt;appropriate viewing utilities on the CD-ROM. For searching, the non-text items are&lt;br&gt;represented by textual descriptions in the MG index.&lt;br&gt;
A request which requires some computation on the server, such as the submission of a&lt;br&gt;query from a user, would normally be handled with CGI requests.  On our system,&lt;br&gt;such requests are invoked by URL’s starting      &lt;br&gt;
http://127.0.0.1/server/   .  These are&lt;br&gt;
internally routed to handler routines within the server itself – particularly to MG&lt;br&gt;components.&lt;br&gt;
The major implementation difficulty experienced was with the IP network software, on&lt;br&gt;machines which did not have network cards or modem software.  To avoid installation&lt;br&gt;complexity we chose to implement our own network layer to be used on such&lt;br&gt;machines.  In the absence of networking software the server loads our internal network&lt;br&gt;software and communicates using that.&lt;br&gt;
&lt;b&gt;3 . Searching and navigating a collection&lt;/b&gt;&lt;br&gt;
The primary access method for documents in the United Nations University collection&lt;br&gt;is keyword search (Figure 2a).  The system supports searching over the &lt;i&gt;full&lt;/i&gt; text of the&lt;br&gt;document—not merely a document surrogate as is common in many commercial&lt;br&gt;retrieval systems.  While other collections we have built support a syntax for full&lt;br&gt;Boolean searching, early user feedback from a similar document set (the Humanitarian&lt;br&gt;Development collection, put together by the Global Help Project) indicated that Boolean&lt;br&gt;searching was more confusing than helpful for the targeted users. Previous research&lt;br&gt;suggests that difficulties with Boolean syntax and semantics are common, and are&lt;br&gt;observed in diverse user groups (Borgman, 1996; Greene et al, 1990). Transaction log&lt;br&gt;analysis over a number of library retrieval systems indicates that the most popular&lt;br&gt;Boolean operator by far is the AND, with the Boolean OR and NOT rarely present in&lt;br&gt;queries (Peters, 1993); we have confirmed this result in another New Zealand Digital&lt;br&gt;Library collection (Jones et al, 1998). For all these reasons, the United Nations&lt;br&gt;University interface default is ranked retrieval. However, to enable users to construct&lt;br&gt;high-precision Boolean AND searches where necessary, selecting “search
for ALL&lt;br&gt;the words” in the querying string produces the syntax-free equivalent of an AND query.&lt;br&gt;
     &lt;br&gt;
Figure 2:  (a) Initial search screen for the UNU collection and (b)  search preferences&lt;br&gt;
page&lt;br&gt;
By default, search terms are stemmed and case differences are ignored.  Most&lt;br&gt;transaction log analysis from library online catalogs, digital libraries, and WWW search&lt;br&gt;engines indicates that users tend to submit extremely brief queries.  For example, the&lt;br&gt;average query length for the New Zealand Digital Library’s &lt;i&gt;Computer Science&lt;br&gt;Technical Report&lt;/i&gt; collection is only 2.5 words (Jones et al, 1998), a typical figure&lt;br&gt;mirrored in retrieval studies conducted over two decades (Sandore, 1993).  With such&lt;br&gt;
&lt;hr&gt;
&lt;A name=5&gt;&lt;/a&gt;brief queries the major difficulty encountered with search results is low search&lt;br&gt;recall—hence the system automatically expands the query through stemming and case&lt;br&gt;folding.  These defaults can be modified by&lt;br&gt;
The initial search screen  (Figure 2a) also permits users to specify the “granularity” at&lt;br&gt;which their search is done (that is, the size of the text against which the query is&lt;br&gt;matched). Choices include &lt;i&gt;title&lt;/i&gt;, &lt;i&gt;paragraph&lt;/i&gt;, &lt;i&gt;same chapter or section&lt;/i&gt;, and &lt;i&gt;book&lt;/i&gt;. By&lt;br&gt;selecting the smaller passage sizes, users can achieve a greater search precision, while&lt;br&gt;selecting the larger ones tends to give a higher recall.  Regardless of granularity, the&lt;br&gt;results are always displayed in terms of a complete book, opened at the appropriate&lt;br&gt;place.&lt;br&gt;
Figure 3:  Query results page&lt;br&gt;
We support browsing by taking advantage of the fact that the hierarchical structure of&lt;br&gt;United Nations University Press documents is marked up in the document files.  When&lt;br&gt;an item in the “query results” list is selected (Figure 3), the user is presented with a&lt;br&gt;photograph of the document’s front cover and a table of contents with an arrow&lt;br&gt;marking the item’s position in the contents (Figure 4). Folders can be clicked open or&lt;br&gt;closed, allowing the user to travel up and down the document’s structure (in Figure 5,&lt;br&gt;moving from a report up to the section headings for that issue of the bulletin).  Clicking&lt;br&gt;on “expand contents” will expand out the whole table of contents so that the user can&lt;br&gt;browse the titles of all chapters and subsections to get a detailed view of the entire&lt;br&gt;contents.  “Expand text” displays the whole text of the current section or book, which is&lt;br&gt;particularly useful when printing a complete work.&lt;br&gt;
Figure 4:  Viewing a selected item in the query results list&lt;br&gt;
&lt;hr&gt;
&lt;A name=6&gt;&lt;/a&gt;Figure 5:  Moving up the document structure hierarchy&lt;br&gt;
Browsing or searching by subject is supported by  clicking the “subjects” button on the&lt;br&gt;menu options bar of any search or results page . This brings up a list of subjects,&lt;br&gt;represented by bookshelves (Figure 6). Users can click on any bookshelf to look at&lt;br&gt;books on that subject, and click on a book to read it. Similarly, clicking on the “titles”&lt;br&gt;button allows the user to browse through an alphabetized list of titles. If the user is&lt;br&gt;currently viewing a document when the “subjects” or “titles” button is clicked, s/he will&lt;br&gt;be taken to the place in the subjects or titles list that corresponds to that book. This&lt;br&gt;supports the user in browsing for books on the same subject, or for books with similar&lt;br&gt;titles.&lt;br&gt;
Figure 6:  Browsing by subject&lt;br&gt;
&lt;b&gt;4 . Conclusions&lt;/b&gt;&lt;br&gt;
Despite near-universal current practice, the World-Wide Web is by no means the only&lt;br&gt;way to deliver digital library services.  Local networks and CD-ROM disks can be a&lt;br&gt;viable alternative—and a necessary one in many operating environments. The humble&lt;br&gt;CD-ROM can hold a lot of text, and DVD disks will enable easy distribution of very&lt;br&gt;substantial collections&lt;br&gt;
The challenge is to produce a scheme which can be used for distribution over each of&lt;br&gt;these media, and look just the same to the user. The Greenstone software allows&lt;br&gt;information to be made available in precisely the same form, using precisely the same&lt;br&gt;interface, on a single-user (PC) computer, a local intranet, or the World-Wide Web.&lt;br&gt;One reason for developing this technology was to permit access to important&lt;br&gt;information in the Third World, which runs the risk of falling further behind because of&lt;br&gt;inadequate network access. However, all who find the Internet capricious in terms of&lt;br&gt;remote site availability, and suffer from highly variable and unpredictable network&lt;br&gt;delays, will appreciate the advantages of having digital library information on&lt;br&gt;site—whether in single-user or shared mode.&lt;br&gt;
&lt;hr&gt;
&lt;A name=7&gt;&lt;/a&gt;The United Nations University collection that we have described and illustrated is&lt;br&gt;designed not, as most digital libraries seem to be, for technophiles, but for ordinary&lt;br&gt;people with little or no computer experience. We have again run counter to common&lt;br&gt;practice here to make the interface plain and easy to use. In a quest to improve usability&lt;br&gt;for the ordinary person we have sacrificed features—actually deleted them from our&lt;br&gt;software—that, although powerful, we have observed to be rarely employed by real&lt;br&gt;users answering their real information needs.&lt;br&gt;
&lt;b&gt;References&lt;/b&gt;&lt;br&gt;
Borgman, C.L. (1996) Why are online catalogs still hard to use&lt;i&gt;?  Journal of the&lt;/i&gt;&lt;br&gt;
&lt;i&gt;American Society for Information Science&lt;/i&gt; 47(7), pp. 493-503.&lt;br&gt;
Chowdhury, G.G. (1996) Developing modern information systems and services:&lt;br&gt;
Africa’s challenges for the future, &lt;i&gt;Online &amp;amp; CDROM Review&lt;/i&gt; 20(3), pp. 145-&lt;br&gt;146.&lt;br&gt;
El-Hadidy, B. (1994)  The breakeven point for using CD-ROM versus online:  a case&lt;br&gt;
study for database access in a developing country, &lt;i&gt;Journal of the American&lt;br&gt;Society for Information Science&lt;/i&gt; 45(4), pp. 273-283.&lt;br&gt;
Greene, S.L., Devlin, S.J., (1990) Cannata, P.E., and Gomez, L.M. No Ifs, ANDs or&lt;br&gt;
Ors:  a study of database querying, &lt;i&gt;International Journal of Man-Machine&lt;br&gt;Studies&lt;/i&gt; 32(3), pp. 303-326.&lt;br&gt;
Holmes, G., and Rogers, W.J. (197) Gathering and indexing rich fragments of the&lt;br&gt;
World-Wide Web, &lt;i&gt;Proceedings of the International Conference on Computers&lt;br&gt;in Education 1997&lt;/i&gt; (Sarawak, Malaysia, Dec. 2-6), pp. 554-562.&lt;br&gt;
Jones, S., Cunningham, S.J., and McNab, R. (1998) An analysis of usage of a digital&lt;br&gt;
library, &lt;i&gt;Working Paper 98/13&lt;/i&gt;, Department of Computer Science, University of&lt;br&gt;Waikato (Hamilton, New Zealand.&lt;br&gt;
Peters, T. (1993) The history and development of transaction log analysis, &lt;i&gt;Library Hi-&lt;/i&gt;&lt;br&gt;
&lt;i&gt;Tech &lt;/i&gt;11(2), pp. 41-66.&lt;br&gt;
Sandore, B. (1993) Applying the results of transaction log analysis, &lt;i&gt;Library Hi-Tech&lt;/i&gt;&lt;br&gt;
11(2), pp. 87-97.&lt;br&gt;
Wall, L., Christiansen, T., and Schwartz, R.L. (1996) &lt;i&gt; Programming Perl.&lt;/i&gt; O’Reilly,&lt;br&gt;
Sebastopol (CA, USA).&lt;br&gt;
White, W.D. (1992) CD-ROM in developing countries, &lt;i&gt;CD-ROM Professional&lt;/i&gt; (May),&lt;br&gt;
pp. 32-35.&lt;br&gt;
Witten, I.H., Moffat, A., and Bell, T.C. (1994) &lt;i&gt;Managing Gigabytes&lt;/i&gt;. Van Nostrand&lt;br&gt;
Reinhold, New York, New York.&lt;br&gt;
&lt;hr&gt;


</Content>
</Section>
</Archive>