1 <chapter id="introduction">
2 <title>Introduction</title>
4 <section id="overview">
5 <title>Overview</title>
8 &zebra; is a free, fast, friendly information management system. It can
9 index records in &acro.xml;/&acro.sgml;, &acro.marc;, e-mail archives and many other
10 formats, and quickly find them using a combination of boolean
11 searching and relevance ranking. Search-and-retrieve applications can
12 be written using &acro.api;s in a wide variety of languages, communicating
13 with the &zebra; server using industry-standard information-retrieval
14 protocols or web services.
17 &zebra; is licensed Open Source, and can be
18 deployed by anyone for any purpose without license fees. The C source
19 code is open to anybody to read and change under the GPL license.
22 &zebra; is a networked component which acts as a
23 reliable &acro.z3950; server
24 for both record/document search, presentation, insert, update and
25 delete operations. In addition, it understands the &acro.sru; family of
26 webservices, which exist in &acro.rest; &acro.get;/&acro.post; and truly
30 &zebra; is available as MS Windows 2003 Server (32 bit) self-extracting
31 package as well as GNU/Debian Linux (32 bit and 64 bit) precompiled
32 packages. It has been deployed successfully on other Unix systems,
33 including Sun Sparc, HP Unix, and many variants of Linux and BSD
37 <ulink url="http://www.indexdata.com/zebra/">http://www.indexdata.com/zebra/</ulink>
38 <ulink url="http://ftp.indexdata.dk/pub/zebra/win32/">http://ftp.indexdata.dk/pub/zebra/win32/</ulink>
39 <ulink url="http://ftp.indexdata.dk/pub/zebra/debian/">http://ftp.indexdata.dk/pub/zebra/debian/</ulink>
43 <ulink url="http://indexdata.dk/zebra/">&zebra;</ulink>
44 is a high-performance, general-purpose structured text
45 indexing and retrieval engine. It reads records in a
46 variety of input formats (e.g. email, &acro.xml;, &acro.marc;) and provides access
47 to them through a powerful combination of boolean search
48 expressions and relevance-ranked free-text queries.
52 &zebra; supports large databases (tens of millions of records,
53 tens of gigabytes of data). It allows safe, incremental
54 database updates on live systems. Because &zebra; supports
55 the industry-standard information retrieval protocol, &acro.z3950;,
56 you can search &zebra; databases using an enormous variety of
57 programs and toolkits, both commercial and free, which understand
58 this protocol. Application libraries are available to allow
59 bespoke clients to be written in Perl, C, C++, Java, Tcl, Visual
60 Basic, Python, &acro.php; and more - see the
61 <ulink url="&url.zoom;">&acro.zoom; web site</ulink>
62 for more information on some of these client toolkits.
66 This document is an introduction to the &zebra; system. It explains
67 how to compile the software, how to prepare your first database,
68 and how to configure the server to give you the
69 functionality that you need.
73 <section id="features">
74 <title>&zebra; Features Overview</title>
82 <entry><xref linkend=""/></entry>
88 <entry><xref linkend=""/></entry>
94 <entry><xref linkend=""/></entry>
99 <section id="features-document">
100 <title>&zebra; Document Model</title>
102 <table id="table-features-document" frame="top">
103 <title>&zebra; document model</title>
105 <colspec colwidth="1*" colname="feature"/>
106 <colspec colwidth="1*" colname="availability"/>
107 <colspec colwidth="3*" colname="notes"/>
108 <colspec colwidth="2*" colname="references"/>
111 <entry>Feature</entry>
112 <entry>Availability</entry>
114 <entry>Reference</entry>
119 <entry>Complex semi-structured Documents</entry>
120 <entry>&acro.xml; and &acro.grs1; Documents</entry>
121 <entry>Both &acro.xml; and &acro.grs1; documents exhibit a &acro.dom; like internal
122 representation allowing for complex indexing and display rules</entry>
123 <entry><xref linkend="record-model-alvisxslt"/> and
124 <xref linkend="grs"/></entry>
127 <entry>Input document formats</entry>
128 <entry>&acro.xml;, &acro.sgml;, Text, ISO2709 (&acro.marc;)</entry>
130 A system of input filters driven by
131 regular expressions allows most ASCII-based
132 data formats to be easily processed.
133 &acro.sgml;, &acro.xml;, ISO2709 (&acro.marc;), and raw text are also
135 <entry><xref linkend="componentmodules"/></entry>
138 <entry>Document storage</entry>
139 <entry>Index-only, Key storage, Document storage</entry>
140 <entry>Data can be, and usually is, imported
141 into &zebra;'s own storage, but &zebra; can also refer to
142 external files, building and maintaining indexes of "live"
152 <section id="features-search">
153 <title>&zebra; Search Features</title>
155 <table id="table-features-search" frame="top">
156 <title>&zebra; search functionality</title>
158 <colspec colwidth="1*" colname="feature"/>
159 <colspec colwidth="1*" colname="availability"/>
160 <colspec colwidth="3*" colname="notes"/>
161 <colspec colwidth="2*" colname="references"/>
164 <entry>Feature</entry>
165 <entry>Availability</entry>
167 <entry>Reference</entry>
172 <entry>Query languages</entry>
173 <entry>&acro.cql; and &acro.rpn;/&acro.pqf;</entry>
174 <entry>The type-1 Reverse Polish Notation (&acro.rpn;)
175 and its textual representation Prefix Query Format (&acro.pqf;) are
176 supported. The Common Query Language (&acro.cql;) can be configured as
177 a mapping from &acro.cql; to &acro.rpn;/&acro.pqf;</entry>
178 <entry><xref linkend="querymodel-query-languages-pqf"/> and
179 <xref linkend="querymodel-cql-to-pqf"/></entry>
182 <entry>Complex boolean query tree</entry>
183 <entry>&acro.cql; and &acro.rpn;/&acro.pqf;</entry>
184 <entry>Both &acro.cql; and &acro.rpn;/&acro.pqf; allow atomic query parts (&acro.apt;) to
185 be combined into complex boolean query trees</entry>
186 <entry><xref linkend="querymodel-rpn-tree"/></entry>
189 <entry>Field search</entry>
190 <entry>user defined</entry>
191 <entry>Atomic query parts (&acro.apt;) are either general, or
192 directed at user-specified document fields
194 <entry><xref linkend="querymodel-atomic-queries"/>,
195 <xref linkend="querymodel-use-string"/>,
196 <xref linkend="querymodel-bib1-use"/>, and
197 <xref linkend="querymodel-idxpath-use"/></entry>
200 <entry>Data normalization</entry>
201 <entry>user defined</entry>
202 <entry>Data normalization, text tokenization and character
203 mappings can be applied during indexing and searching</entry>
204 <entry><xref linkend="fields-and-charsets"/></entry>
207 <entry>Predefined field types</entry>
208 <entry>user defined</entry>
209 <entry>Data fields can be indexed as phrase, as into word
210 tokenized text, as numeric values, URLs, dates, and raw binary
212 <entry><xref linkend="character-map-files"/> and
213 <xref linkend="querymodel-pqf-apt-mapping-structuretype"/>
217 <entry>Regular expression matching</entry>
218 <entry>available</entry>
219 <entry>Full regular expression matching and "approximate
220 matching" (e.g. spelling mistake corrections) are handled.</entry>
221 <entry><xref linkend="querymodel-regular"/></entry>
224 <entry>Term truncation</entry>
225 <entry>left, right, left-and-right</entry>
226 <entry>The truncation attribute specifies whether variations of
227 one or more characters are allowed between search term and hit
228 terms, or not. Using non-default truncation attributes will
229 broaden the document hit set of a search query.</entry>
230 <entry><xref linkend="querymodel-bib1-truncation"/></entry>
233 <entry>Fuzzy searches</entry>
234 <entry>Spelling correction</entry>
235 <entry>In addition, fuzzy searches are implemented, where one
236 spelling mistake in search terms is matched</entry>
237 <entry><xref linkend="querymodel-bib1-truncation"/></entry>
244 <section id="features-scan">
245 <title>&zebra; Index Scanning</title>
247 <table id="table-features-scan" frame="top">
248 <title>&zebra; index scanning</title>
250 <colspec colwidth="1*" colname="feature"/>
251 <colspec colwidth="1*" colname="availability"/>
252 <colspec colwidth="3*" colname="notes"/>
253 <colspec colwidth="2*" colname="references"/>
256 <entry>Feature</entry>
257 <entry>Availability</entry>
259 <entry>Reference</entry>
265 <entry>term suggestions</entry>
266 <entry><literal>Scan</literal> on a given named index returns all the
267 indexed terms in lexicographical order near the given start
268 term. This can be used to create drop-down menus and search
270 <entry><xref linkend="querymodel-operation-type-scan"/> and
271 <xref linkend="querymodel-atomic-queries"/>
275 <entry>Facetted browsing</entry>
276 <entry>available</entry>
277 <entry>Zebra 2.1 and allows retrieval of facets for
280 <entry><xref linkend="querymodel-zebra-attr-scan"/></entry>
283 <entry>Drill-down or refine-search</entry>
284 <entry>partially</entry>
285 <entry>scanning in result sets can be used to implement
286 drill-down in search clients</entry>
287 <entry><xref linkend="querymodel-zebra-attr-scan"/></entry>
294 <section id="features-presentation">
295 <title>&zebra; Document Presentation</title>
297 <table id="table-features-presentation" frame="top">
298 <title>&zebra; document presentation</title>
300 <colspec colwidth="1*" colname="feature"/>
301 <colspec colwidth="1*" colname="availability"/>
302 <colspec colwidth="3*" colname="notes"/>
303 <colspec colwidth="2*" colname="references"/>
306 <entry>Feature</entry>
307 <entry>Availability</entry>
309 <entry>Reference</entry>
314 <entry>Hit count</entry>
316 <entry>Search results include at any time the total hit count of a given
317 query, either exact computed, or approximative, in case that the
318 hit count exceeds a possible pre-defined hit set truncation
321 <xref linkend="querymodel-zebra-local-attr-limit"/> and
322 <xref linkend="zebra-cfg"/>
326 <entry>Paged result sets</entry>
328 <entry>Paging of search requests and present/display request
329 can return any successive number of records from any start
330 position in the hit set, i.e. it is trivial to provide search
331 results in successive pages of any size.</entry>
335 <entry>&acro.xml; document transformations</entry>
336 <entry>&acro.xslt; based</entry>
337 <entry> Record presentation can be performed in many
338 pre-defined &acro.xml; data
339 formats, where the original &acro.xml; records are on-the-fly transformed
340 through any preconfigured &acro.xslt; transformation. It is therefore
341 trivial to present records in short/full &acro.xml; views, transforming to
342 RSS, Dublin Core, or other &acro.xml; based data formats, or transform
343 records to XHTML snippets ready for inserting in XHTML pages.</entry>
345 <xref linkend="record-model-alvisxslt-elementset"/></entry>
348 <entry>Binary record transformations</entry>
349 <entry>&acro.marc;, &acro.usmarc;, &acro.marc21; and &acro.marcxml;</entry>
350 <entry>post-filter record transformations</entry>
354 <entry>Record Syntaxes</entry>
356 <entry> Multiple record syntaxes
357 for data retrieval: &acro.grs1;, &acro.sutrs;,
358 &acro.xml;, ISO2709 (&acro.marc;), etc. Records can be mapped between
359 record syntaxes and schemas on the fly.</entry>
363 <entry>&zebra; internal metadata</entry>
365 <entry> &zebra; internal document metadata can be fetched in
366 &acro.sutrs; and &acro.xml; record syntaxes. Those are useful in client
367 applications.</entry>
368 <entry><xref linkend="special-retrieval"/></entry>
371 <entry>&zebra; internal raw record data</entry>
373 <entry> &zebra; internal raw, binary record data can be fetched in
374 &acro.sutrs; and &acro.xml; record syntaxes, leveraging %zebra; to a
375 binary storage system</entry>
376 <entry><xref linkend="special-retrieval"/></entry>
379 <entry>&zebra; internal record field data</entry>
381 <entry> &zebra; internal record field data can be fetched in
382 &acro.sutrs; and &acro.xml; record syntaxes. This makes very fast minimal
383 record data displays possible.</entry>
384 <entry><xref linkend="special-retrieval"/></entry>
391 <section id="features-sort-rank">
392 <title>&zebra; Sorting and Ranking</title>
394 <table id="table-features-sort-rank" frame="top">
395 <title>&zebra; sorting and ranking</title>
397 <colspec colwidth="1*" colname="feature"/>
398 <colspec colwidth="1*" colname="availability"/>
399 <colspec colwidth="3*" colname="notes"/>
400 <colspec colwidth="2*" colname="references"/>
403 <entry>Feature</entry>
404 <entry>Availability</entry>
406 <entry>Reference</entry>
412 <entry>numeric, lexicographic</entry>
413 <entry>Sorting on the basis of alpha-numeric and numeric data
414 is supported. Alphanumeric sorts can be configured for
415 different data encodings and locales for European languages.</entry>
416 <entry><xref linkend="administration-ranking-sorting"/> and
417 <xref linkend="querymodel-zebra-attr-sorting"/></entry>
420 <entry>Combined sorting</entry>
422 <entry>Sorting on the basis of combined sorts  e.g. combinations of
423 ascending/descending sorts of lexicographical/numeric/date field data
425 <entry><xref linkend="administration-ranking-sorting"/></entry>
428 <entry>Relevance ranking</entry>
429 <entry>TF-IDF like</entry>
430 <entry>Relevance-ranking of free-text queries is supported
431 using a TF-IDF like algorithm.</entry>
432 <entry><xref linkend="administration-ranking-dynamic"/></entry>
435 <entry>Static pre-ranking</entry>
437 <entry>Enables pre-index time ranking of documents where hit
438 lists are ordered first by ascending static rank, then by
439 ascending document ID.</entry>
440 <entry><xref linkend="administration-ranking-static"/></entry>
448 <section id="features-updates">
449 <title>&zebra; Live Updates</title>
452 <table id="table-features-updates" frame="top">
453 <title>&zebra; live updates</title>
455 <colspec colwidth="1*" colname="feature"/>
456 <colspec colwidth="1*" colname="availability"/>
457 <colspec colwidth="3*" colname="notes"/>
458 <colspec colwidth="2*" colname="references"/>
461 <entry>Feature</entry>
462 <entry>Availability</entry>
464 <entry>Reference</entry>
469 <entry>Incremental and batch updates</entry>
471 <entry>It is possible to schedule record inserts/updates/deletes in any
472 quantity, from single individual handled records to batch updates
473 in strikes of any size, as well as total re-indexing of all records
474 from file system. </entry>
475 <entry><xref linkend="zebraidx"/></entry>
478 <entry>Remote updates</entry>
479 <entry>&acro.z3950; extended services</entry>
480 <entry>Updates can be performed from remote locations using the
481 &acro.z3950; extended services. Access to extended services can be
482 login-password protected.</entry>
483 <entry><xref linkend="administration-extended-services"/> and
484 <xref linkend="zebra-cfg"/></entry>
487 <entry>Live updates</entry>
488 <entry>transaction based</entry>
489 <entry> Data updates are transaction based and can be performed
490 on running &zebra; systems. Full searchability is preserved
491 during life data update due to use of shadow disk areas for
492 update operations. Multiple update transactions at the same
493 time are lined up, to be performed one after each other. Data
494 integrity is preserved.</entry>
495 <entry><xref linkend="shadow-registers"/></entry>
502 <section id="features-protocol">
503 <title>&zebra; Networked Protocols</title>
505 <table id="table-features-protocol" frame="top">
506 <title>&zebra; networked protocols</title>
508 <colspec colwidth="1*" colname="feature"/>
509 <colspec colwidth="1*" colname="availability"/>
510 <colspec colwidth="3*" colname="notes"/>
511 <colspec colwidth="2*" colname="references"/>
514 <entry>Feature</entry>
515 <entry>Availability</entry>
517 <entry>Reference</entry>
522 <entry>Fundamental operations</entry>
523 <entry>&acro.z3950;/&acro.sru; <literal>explain</literal>,
524 <literal>search</literal>, <literal>scan</literal>, and
525 <literal>update</literal></entry>
527 <entry><xref linkend="querymodel-operation-types"/></entry>
530 <entry>&acro.z3950; protocol support</entry>
532 <entry> Protocol facilities supported are:
533 <literal>init</literal>, <literal>search</literal>,
534 <literal>present</literal> (retrieval),
535 Segmentation (support for very large records),
536 <literal>delete</literal>, <literal>scan</literal>
537 (index browsing), <literal>sort</literal>,
538 <literal>close</literal> and support for the <literal>update</literal>
539 Extended Service to add or replace an existing &acro.xml;
540 record. Piggy-backed presents are honored in the search
541 request. Named result sets are supported.</entry>
542 <entry><xref linkend="protocol-support"/></entry>
545 <entry>Web Service support</entry>
546 <entry>&acro.sru;</entry>
547 <entry> The protocol operations <literal>explain</literal>,
548 <literal>searchRetrieve</literal> and <literal>scan</literal>
549 are supported. <ulink url="&url.cql;">&acro.cql;</ulink> to internal
550 query model &acro.rpn;
551 conversion is supported. Extended RPN queries
552 for search/retrieve and scan are supported.</entry>
553 <entry><xref linkend="zebrasrv-sru-support"/></entry>
560 <section id="features-scalability">
561 <title>&zebra; Data Size and Scalability</title>
563 <table id="table-features-scalability" frame="top">
564 <title>&zebra; data size and scalability</title>
566 <colspec colwidth="1*" colname="feature"/>
567 <colspec colwidth="1*" colname="availability"/>
568 <colspec colwidth="3*" colname="notes"/>
569 <colspec colwidth="2*" colname="references"/>
572 <entry>Feature</entry>
573 <entry>Availability</entry>
575 <entry>Reference</entry>
580 <entry>No of records</entry>
581 <entry>40-60 million</entry>
586 <entry>Data size</entry>
587 <entry>100 GB of record data</entry>
588 <entry>&zebra; based applications have successfully indexed up
589 to 100 GB of record data</entry>
593 <entry>Scale out</entry>
594 <entry>multiple discs</entry>
599 <entry>Performance</entry>
600 <entry><literal>O(n * log N)</literal></entry>
601 <entry> &zebra; query speed and performance is affected roughly by
602 <literal>O(log N)</literal>,
603 where <literal>N</literal> is the total database size, and by
604 <literal>O(n)</literal>, where <literal>n</literal> is the
605 specific query hit set size.</entry>
609 <entry>Average search times</entry>
611 <entry> Even on very large size databases hit rates of 20 queries per
612 seconds with average query answering time of 1 second are possible,
613 provided that the boolean queries are constructed sufficiently
614 precise to result in hit sets of the order of 1000 to 5.000
619 <entry>Large databases</entry>
620 <entry>64 bit file pointers</entry>
621 <entry>64 file pointers assure that register files can extend
622 the 2 GB limit. Logical files can be
623 automatically partitioned over multiple disks, thus allowing for
624 large databases.</entry>
632 <section id="features-platforms">
633 <title>&zebra; Supported Platforms</title>
635 <table id="table-features-platforms" frame="top">
636 <title>&zebra; supported platforms</title>
638 <colspec colwidth="1*" colname="feature"/>
639 <colspec colwidth="1*" colname="availability"/>
640 <colspec colwidth="3*" colname="notes"/>
641 <colspec colwidth="2*" colname="references"/>
644 <entry>Feature</entry>
645 <entry>Availability</entry>
647 <entry>Reference</entry>
654 <entry>GNU Linux (32 and 64bit), journaling Reiser or (better)
656 on disks. NFS file systems are not supported.
657 GNU/Debian Linux packages are available</entry>
658 <entry><xref linkend="installation-debian"/></entry>
662 <entry>tar-ball</entry>
663 <entry>&zebra; is written in portable C, so it runs on most
665 Usual tar-ball install possible on many major Unix systems</entry>
666 <entry><xref linkend="installation-unix"/></entry>
669 <entry>Windows</entry>
670 <entry>NT/2000/2003/XP</entry>
671 <entry>&zebra; runs as well on Windows (NT/2000/2003/XP).
672 Windows installer packages available</entry>
673 <entry><xref linkend="installation-win32"/></entry>
683 <section id="introduction-apps">
684 <title>References and &zebra; based Applications</title>
686 &zebra; has been deployed in numerous applications, in both the
687 academic and commercial worlds, in application domains as diverse
688 as bibliographic catalogues, Geo-spatial information, structured
689 vocabulary browsing, government information locators, civic
690 information systems, environmental observations, museum information
694 Notable applications include the following:
698 <section id="koha-ils">
699 <title>Koha free open-source ILS</title>
701 <ulink url="http://www.koha.org/">Koha</ulink> is a full-featured
702 open-source ILS, initially developed in
703 New Zealand by Katipo Communications Ltd, and first deployed in
704 January of 2000 for Horowhenua Library Trust. It is currently
705 maintained by a team of software providers and library technology
706 staff from around the globe.
709 <ulink url="http://liblime.com/">LibLime</ulink>,
710 a company that is marketing and supporting Koha, adds in
711 the new release of Koha 3.0 the &zebra;
712 database server to drive its bibliographic database.
715 In early 2005, the Koha project development team began looking at
716 ways to improve &acro.marc; support and overcome scalability limitations
717 in the Koha 2.x series. After extensive evaluations of the best
718 of the Open Source textual database engines - including MySQL
719 full-text searching, PostgreSQL, Lucene and Plucene - the team
723 "&zebra; completely eliminates scalability limitations, because it
724 can support tens of millions of records." explained Joshua
725 Ferraro, LibLime's Technology President and Koha's Project
726 Release Manager. "Our performance tests showed search results in
727 under a second for databases with over 5 million records on a
728 modest i386 900Mhz test server."
731 "&zebra; also includes support for true boolean search expressions
732 and relevance-ranked free-text queries, both of which the Koha
733 2.x series lack. &zebra; also supports incremental and safe
734 database updates, which allow on-the-fly record
735 management. Finally, since &zebra; has at its heart the &acro.z3950;
736 protocol, it greatly improves Koha's support for that critical
740 Although the bibliographic database will be moved to &zebra;, Koha
741 3.0 will continue to use a relational SQL-based database design
742 for the 'factual' database. "Relational database managers have
743 their strengths, in spite of their inability to handle large
744 numbers of bibliographic records efficiently," summed up Ferraro,
745 "We're taking the best from both worlds in our redesigned Koha
749 See also LibLime's newsletter article
750 <ulink url="http://www.liblime.com/newsletter/2006/01/features/koha-earns-its-stripes/">
751 Koha Earns its Stripes</ulink>.
756 <section id="kete-dom">
757 <title>Kete Open Source Digital Library and Archiving software</title>
759 <ulink url="http://kete.net.nz/">Kete</ulink> is a digital object
760 management repository, initially developed in
761 New Zealand. Initial development has
762 been a partnership between the Horowhenua Library Trust and
763 Katipo Communications Ltd. funded as part of the Community
764 Partnership Fund in 2006.
765 Kete is purpose built
766 software to enable communities to build their own digital
767 libraries, archives and repositories.
770 It is based on Ruby-on-Rails and MySQL, and integrates the &zebra; server
771 and the &yaz; toolkit for indexing and retrieval of it's content.
772 Zebra is run as separate computer process from the Kete
776 url="http://kete.net.nz/documentation/topics/show/139-managing-zebra">manages
780 Why does Kete wants to use Zebra?? Speed, Scalability and easy
781 integration with Koha. Read their
783 url="http://kete.net.nz/blog/topics/show/44-who-what-why-when-answering-some-of-the-niggly-development-questions">detailed
784 reasoning here.</ulink>
788 <section id="emilda-ils">
789 <title>Emilda open source ILS</title>
791 <ulink url="http://www.emilda.org/">Emilda</ulink>
792 is a complete Integrated Library System, released under the
793 GNU General Public License. It has a
794 full featured Web-OPAC, allowing comprehensive system management
795 from virtually any computer with an Internet connection, has
796 template based layout allowing anyone to alter the visual
797 appearance of Emilda, and is
798 &acro.xml; based language for fast and easy portability to virtually any
800 Currently, Emilda is used at three schools in Espoo, Finland.
803 As a surplus, 100% &acro.marc; compatibility has been achieved using the
804 &zebra; Server from Index Data as backend server.
808 <section id="reindex-ils">
809 <title>ReIndex.Net web based ILS</title>
811 <ulink url="http://www.reindex.net/index.php?lang=en">Reindex.net</ulink>
812 is a netbased library service offering all
813 traditional functions on a very high level plus many new
814 services. Reindex.net is a comprehensive and powerful WEB system
815 based on standards such as &acro.xml; and &acro.z3950;.
816 updates. Reindex supports &acro.marc21;, dan&acro.marc; eller Dublin Core with
820 Reindex.net runs on GNU/Debian Linux with &zebra; and Simpleserver
822 Data for bibliographic data. The relational database system
823 Sybase 9 &acro.xml; is used for
825 Internally &acro.marcxml; is used for bibliographical records. Update
826 utilizes &acro.z3950; extended services.
830 <section id="dads-article-database">
831 <title>DADS - the DTV Article Database
834 DADS is a huge database of more than ten million records, totalling
835 over ten gigabytes of data. The records are metadata about academic
836 journal articles, primarily scientific; about 10% of these
837 metadata records link to the full text of the articles they
838 describe, a body of about a terabyte of information (although the
839 full text is not indexed.)
842 It allows students and researchers at DTU (Danmarks Tekniske
843 Universitet, the Technical College of Denmark) to find and order
844 articles from multiple databases in a single query. The database
845 contains literature on all engineering subjects. It's available
846 on-line through a web gateway, though currently only to registered
850 More information can be found at
851 <ulink url="http://www.dtv.dk/"/> and
852 <ulink url="http://dads.dtv.dk"/>
856 <section id="infonet-eprints">
857 <title>Infonet Eprints</title>
859 The InfoNet Eprints service from the
860 <ulink url="http://www.dtv.dk/">
861 Technical Knowledge Center of Denmark</ulink>
862 provides access to documents stored in
863 eprint/preprint servers and institutional research archives around
864 the world. The service is based on Open Archives Initiative metadata
865 harvesting of selected scientific archives around the world. These
866 open archives offer free and unrestricted access to their contents.
869 Infonet Eprints currently holds 1.4 million records from 16 archives.
870 The online search facility is found at
871 <ulink url="http://preprints.cvt.dk"/>.
875 <section id="alvis-project">
878 The <ulink url="http://www.alvis.info/alvis/">Alvis</ulink> EU
879 project run under the 6th Framework (IST-1-002068-STP)
880 is building a semantic-based peer-to-peer search engine. A
881 consortium of eleven partners from six different European
882 Community countries plus Switzerland and China contribute
883 with expertise in a broad range of specialties including network
884 topologies, routing algorithms, linguistic analysis and
888 The &zebra; information retrieval indexing machine is used inside
889 the Alvis framework to
890 manage huge collections of natural language processed and
891 enhanced &acro.xml; data, coming from a topic relevant web crawl.
892 In this application, &zebra; swallows and manages 37GB of &acro.xml; data
893 in about 4 hours, resulting in search times of fractions of
900 <title>ULS (Union List of Serials)</title>
903 has created a union catalogue for the periodicals of the
904 twenty-one constituent libraries of the University of London and
905 the University of Westminster
906 (<ulink url="http://www.m25lib.ac.uk/ULS/"/>).
907 They have achieved this using an
908 unusual architecture, which they describe as a
909 ``non-distributed virtual union catalogue''.
912 The member libraries send in data files representing their
913 periodicals, including both brief bibliographic data and summary
914 holdings. Then 21 individual &acro.z3950; targets are created, each
915 using &zebra;, and all mounted on the single hardware server.
916 The live service provides a web gateway allowing &acro.z3950; searching
917 of all of the targets or a selection of them. &zebra;'s small
918 footprint allows a relatively modest system to comfortably host
922 More information can be found at
923 <ulink url="http://www.m25lib.ac.uk/ULS/"/>
928 <title>NLI-&acro.z3950; - a Natural Language Interface for Libraries</title>
930 Fernuniversität Hagen in Germany have developed a natural
931 language interface for access to library databases.
933 url="http://ki212.fernuni-hagen.de/nli/NLIintro.html"/> -->
934 In order to evaluate this interface for recall and precision, they
935 chose &zebra; as the basis for retrieval effectiveness. The &zebra;
936 server contains a copy of the GIRT database, consisting of more
937 than 76000 records in &acro.sgml; format (bibliographic records from
938 social science), which are mapped to &acro.marc; for presentation.
941 (GIRT is the German Indexing and Retrieval Testdatabase. It is a
942 standard German-language test database for intelligent indexing
943 and retrieval systems. See
944 <ulink url="http://www.gesis.org/forschung/informationstechnologie/clef-delos.htm"/>)
947 Evaluation will take place as part of the TREC/CLEF campaign 2003
948 <ulink url="http://clef.iei.pi.cnr.it"/>.
949 <!-- or <ulink url="http://www4.eurospider.ch/CLEF/"/> -->
952 For more information, contact Johannes Leveling
953 <email>Johannes.Leveling@FernUni-Hagen.De</email>
957 <section id="various-web-indexes">
958 <title>Various web indexes</title>
960 &zebra; has been used by a variety of institutions to construct
961 indexes of large web sites, typically in the region of tens of
962 millions of pages. In this role, it functions somewhat similarly
963 to the engine of Google or AltaVista, but for a selected intranet
964 or a subset of the whole Web.
967 For example, Liverpool University's web-search facility (see on
969 <ulink url="http://www.liv.ac.uk/"/>
970 and many sub-pages) works by relevance-searching a &zebra; database
971 which is populated by the Harvest-NG web-crawling software.
974 For more information on Liverpool university's intranet search
975 architecture, contact John Gilbertson
976 <email>jgilbert@liverpool.ac.uk</email>
980 has recently modified the Harvest web indexer to use &zebra; as
981 its native repository engine. His comments on the switch over
982 from the old engine are revealing:
985 The first results after some testing with &zebra; are very
986 promising. The tests were done with around 220,000 SOIF files,
987 which occupies 1.6GB of disk space.
990 Building the index from scratch takes around one hour with &zebra;
991 where [old-engine] needs around five hours. While [old-engine]
992 blocks search requests when updating its index, &zebra; can still
993 answer search requests.
995 &zebra; supports incremental indexing which will speed up indexing
999 While the search time of [old-engine] varies from some seconds
1000 to some minutes depending how expensive the query is, &zebra;
1001 usually takes around one to three seconds, even for expensive
1004 &zebra; can search more than 100 times faster than [old-engine]
1005 and can process multiple search requests simultaneously
1008 I am very happy to see such nice software available under GPL.
1016 <section id="introduction-support">
1017 <title>Support</title>
1019 You can get support for &zebra; from at least three sources.
1022 First, there's the &zebra; web site at
1023 <ulink url="&url.idzebra;"/>,
1024 which always has the most recent version available for download.
1025 If you have a problem with &zebra;, the first thing to do is see
1026 whether it's fixed in the current release.
1029 Second, there's the &zebra; mailing list. Its home page at
1030 <ulink url="&url.idzebra.mailinglist;"/>
1031 includes a complete archive of all messages that have ever been
1032 posted on the list. The &zebra; mailing list is used both for
1033 announcements from the authors (new
1034 releases, bug fixes, etc.) and general discussion. You are welcome
1035 to seek support there. Join by filling the form on the list home page.
1039 <!-- Keep this comment at the end of the file
1044 sgml-minimize-attributes:nil
1045 sgml-always-quote-attributes:t
1048 sgml-parent-document: "zebra.xml"
1049 sgml-local-catalogs: nil
1050 sgml-namecase-general:t