1 <chapter id="introduction">
2 <!-- $Id: introduction.xml,v 1.49 2007-02-05 14:32:31 marc Exp $ -->
3 <title>Introduction</title>
5 <section id="overview">
6 <title>Overview</title>
9 &zebra; is a free, fast, friendly information management system. It can
10 index records in &xml;/&sgml;, &marc;, e-mail archives and many other
11 formats, and quickly find them using a combination of boolean
12 searching and relevance ranking. Search-and-retrieve applications can
13 be written using &api;s in a wide variety of languages, communicating
14 with the &zebra; server using industry-standard information-retrieval
15 protocols or web services.
18 &zebra; is licensed Open Source, and can be
19 deployed by anyone for any purpose without license fees. The C source
20 code is open to anybody to read and change under the GPL license.
23 &zebra; is a networked component which acts as a
24 reliable &z3950; server
25 for both record/document search, presentation, insert, update and
26 delete operations. In addition, it understands the &sru; family of
27 webservices, which exist in &rest; &get;/&post; and truly
31 &zebra; is available as MS Windows 2003 Server (32 bit) self-extracting
32 package as well as GNU/Debian Linux (32 bit and 64 bit) precompiled
33 packages. It has been deployed successfully on other Unix systems,
34 including Sun Sparc, HP Unix, and many variants of Linux and BSD
38 <ulink url="http://www.indexdata.com/zebra/">http://www.indexdata.com/zebra/</ulink>
39 <ulink url="http://ftp.indexdata.dk/pub/zebra/win32/">http://ftp.indexdata.dk/pub/zebra/win32/</ulink>
40 <ulink url="http://ftp.indexdata.dk/pub/zebra/debian/">http://ftp.indexdata.dk/pub/zebra/debian/</ulink>
44 <ulink url="http://indexdata.dk/zebra/">&zebra;</ulink>
45 is a high-performance, general-purpose structured text
46 indexing and retrieval engine. It reads records in a
47 variety of input formats (eg. email, &xml;, &marc;) and provides access
48 to them through a powerful combination of boolean search
49 expressions and relevance-ranked free-text queries.
53 &zebra; supports large databases (tens of millions of records,
54 tens of gigabytes of data). It allows safe, incremental
55 database updates on live systems. Because &zebra; supports
56 the industry-standard information retrieval protocol, &z3950;,
57 you can search &zebra; databases using an enormous variety of
58 programs and toolkits, both commercial and free, which understand
59 this protocol. Application libraries are available to allow
60 bespoke clients to be written in Perl, C, C++, Java, Tcl, Visual
61 Basic, Python, &php; and more - see the
62 <ulink url="&url.zoom;">&zoom; web site</ulink>
63 for more information on some of these client toolkits.
67 This document is an introduction to the &zebra; system. It explains
68 how to compile the software, how to prepare your first database,
69 and how to configure the server to give you the
70 functionality that you need.
74 <section id="features">
75 <title>&zebra; Features Overview</title>
83 <entry><xref linkend=""/></entry>
89 <entry><xref linkend=""/></entry>
95 <entry><xref linkend=""/></entry>
100 <section id="features-document">
101 <title>&zebra; Document Model</title>
103 <table id="table-features-document" frame="top">
104 <title>&zebra; document model</title>
106 <colspec colwidth="1*" colname="feature"/>
107 <colspec colwidth="1*" colname="availability"/>
108 <colspec colwidth="3*" colname="notes"/>
109 <colspec colwidth="2*" colname="references"/>
112 <entry>Feature</entry>
113 <entry>Availability</entry>
115 <entry>Reference</entry>
120 <entry>Complex semi-structured Documents</entry>
121 <entry>&xml; and &grs1; Documents</entry>
122 <entry>Both &xml; and &grs1; documents exhibit a &dom; like internal
123 representation allowing for complex indexing and display rules</entry>
124 <entry><xref linkend="record-model-alvisxslt"/> and
125 <xref linkend="grs"/></entry>
128 <entry>Input document formats</entry>
129 <entry>&xml;, &sgml;, Text, ISO2709 (&marc;)</entry>
131 A system of input filters driven by
132 regular expressions allows most ASCII-based
133 data formats to be easily processed.
134 &sgml;, &xml;, ISO2709 (&marc;), and raw text are also
136 <entry><xref linkend="componentmodules"/></entry>
139 <entry>Document storage</entry>
140 <entry>Index-only, Key storage, Document storage</entry>
141 <entry>Data can be, and usually is, imported
142 into &zebra;'s own storage, but &zebra; can also refer to
143 external files, building and maintaining indexes of "live"
153 <section id="features-search">
154 <title>&zebra; Search Features</title>
156 <table id="table-features-search" frame="top">
157 <title>&zebra; search functionality</title>
159 <colspec colwidth="1*" colname="feature"/>
160 <colspec colwidth="1*" colname="availability"/>
161 <colspec colwidth="3*" colname="notes"/>
162 <colspec colwidth="2*" colname="references"/>
165 <entry>Feature</entry>
166 <entry>Availability</entry>
168 <entry>Reference</entry>
173 <entry>Query languages</entry>
174 <entry>&cql; and &rpn;/&pqf;</entry>
175 <entry>The type-1 Reverse Polish Notation (&rpn;)
176 and it's textual representation Prefix Query Format (&pqf;) are
177 supported. The Common Query Language (&cql;) can be configured as
178 a mapping from &cql; to &rpn;/&pqf;</entry>
179 <entry><xref linkend="querymodel-query-languages-pqf"/> and
180 <xref linkend="querymodel-cql-to-pqf"/></entry>
183 <entry>Complex boolean query tree</entry>
184 <entry>&cql; and &rpn;/&pqf;</entry>
185 <entry>Both &cql; and &rpn;/&pqf; allow atomic query parts (&apt;) to
186 be combined into complex boolean query trees</entry>
187 <entry><xref linkend="querymodel-rpn-tree"/></entry>
190 <entry>Field search</entry>
191 <entry>user defined</entry>
192 <entry>Atomic query parts (&apt;) are either general, or
193 directed at user-specified document fields
195 <entry><xref linkend="querymodel-atomic-queries"/>,
196 <xref linkend="querymodel-use-string"/>,
197 <xref linkend="querymodel-bib1-use"/>, and
198 <xref linkend="querymodel-idxpath-use"/></entry>
201 <entry>Data normalization</entry>
202 <entry>user defined</entry>
203 <entry>Data normalization, text tokenization and character
204 mappings can be applied during indexing and searching</entry>
205 <entry><xref linkend="fields-and-charsets"/></entry>
208 <entry>Predefined field types</entry>
209 <entry>user defined</entry>
210 <entry>Data fields can be indexed as phrase, as into word
211 tokenized text, as numeric values, url's, dates, and raw binary
213 <entry><xref linkend="character-map-files"/> and
214 <xref linkend="querymodel-pqf-apt-mapping-structuretype"/>
218 <entry>Regular expression matching</entry>
219 <entry>available</entry>
220 <entry>Full regular expression matching and "approximate
221 matching" (eg. spelling mistake corrections) are handled.</entry>
222 <entry><xref linkend="querymodel-regular"/></entry>
225 <entry>Term truncation</entry>
226 <entry>left, right, left-and-right</entry>
227 <entry>The truncation attribute specifies whether variations of
228 one or more characters are allowed between search term and hit
229 terms, or not. Using non-default truncation attributes will
230 broaden the document hit set of a search query.</entry>
231 <entry><xref linkend="querymodel-bib1-truncation"/></entry>
234 <entry>Fuzzy searches</entry>
235 <entry>Spelling correction</entry>
236 <entry>In addition, fuzzy searches are implemented, where one
237 spelling mistake in search terms is matched</entry>
238 <entry><xref linkend="querymodel-bib1-truncation"/></entry>
245 <section id="features-scan">
246 <title>&zebra; Index Scanning</title>
248 <table id="table-features-scan" frame="top">
249 <title>&zebra; index scanning</title>
251 <colspec colwidth="1*" colname="feature"/>
252 <colspec colwidth="1*" colname="availability"/>
253 <colspec colwidth="3*" colname="notes"/>
254 <colspec colwidth="2*" colname="references"/>
257 <entry>Feature</entry>
258 <entry>Availability</entry>
260 <entry>Reference</entry>
266 <entry>term suggestions</entry>
267 <entry><literal>Scan</literal> on a given named index returns all the
268 indexed terms in lexicographical order near the given start
269 term. This can be used to create drop-down menus and search
271 <entry><xref linkend="querymodel-operation-type-scan"/> and
272 <xref linkend="querymodel-atomic-queries"/>
276 <entry>Facetted browsing</entry>
277 <entry>partial</entry>
278 <entry>&zebra; supports <literal>scan inside a hit
279 set</literal> from a previous search, thus reducing the listed
281 subset of terms found in the documents/records of the hit
283 <entry><xref linkend="querymodel-zebra-attr-scan"/></entry>
286 <entry>Drill-down or refine-search</entry>
287 <entry>partially</entry>
288 <entry>scanning in result sets can be used to implement
289 drill-down in search clients</entry>
290 <entry><xref linkend="querymodel-zebra-attr-scan"/></entry>
297 <section id="features-presentation">
298 <title>&zebra; Document Presentation</title>
300 <table id="table-features-presentation" frame="top">
301 <title>&zebra; document presentation</title>
303 <colspec colwidth="1*" colname="feature"/>
304 <colspec colwidth="1*" colname="availability"/>
305 <colspec colwidth="3*" colname="notes"/>
306 <colspec colwidth="2*" colname="references"/>
309 <entry>Feature</entry>
310 <entry>Availability</entry>
312 <entry>Reference</entry>
317 <entry>Hit count</entry>
319 <entry>Search results include at any time the total hit count of a given
320 query, either exact computed, or approximative, in case that the
321 hit count exceeds a possible pre-defined hit set truncation
324 <xref linkend="querymodel-zebra-local-attr-limit"/> and
325 <xref linkend="zebra-cfg"/>
329 <entry>Paged result sets</entry>
331 <entry>Paging of search requests and present/display request
332 can return any successive number of records from any start
333 position in the hit set, i.e. it is trivial to provide search
334 results in successive pages of any size.</entry>
338 <entry>&xml; document transformations</entry>
339 <entry>&xslt; based</entry>
340 <entry> Record presentation can be performed in many
341 pre-defined &xml; data
342 formats, where the original &xml; records are on-the-fly transformed
343 through any preconfigured &xslt; transformation. It is therefore
344 trivial to present records in short/full &xml; views, transforming to
345 RSS, Dublin Core, or other &xml; based data formats, or transform
346 records to XHTML snippets ready for inserting in XHTML pages.</entry>
348 <xref linkend="record-model-alvisxslt-elementset"/></entry>
351 <entry>Binary record transformations</entry>
352 <entry>&marc;, &usmarc;, &marc21; and &marcxml;</entry>
353 <entry>post-filter record transformations</entry>
357 <entry>Record Syntaxes</entry>
359 <entry> Multiple record syntaxes
360 for data retrieval: &grs1;, &sutrs;,
361 &xml;, ISO2709 (&marc;), etc. Records can be mapped between
362 record syntaxes and schemas on the fly.</entry>
366 <entry>&zebra; internal metadata</entry>
368 <entry> &zebra; internal document metadata can be fetched in
369 &sutrs; and &xml; record syntaxes. Those are useful in client
370 applications.</entry>
371 <entry><xref linkend="special-retrieval"/></entry>
374 <entry>&zebra; internal raw record data</entry>
376 <entry> &zebra; internal raw, binary record data can be fetched in
377 &sutrs; and &xml; record syntaxes, leveraging %zebra; to a
378 binary storage system</entry>
379 <entry><xref linkend="special-retrieval"/></entry>
382 <entry>&zebra; internal record field data</entry>
384 <entry> &zebra; internal record field data can be fetched in
385 &sutrs; and &xml; record syntaxes. This makes very fast minimal
386 record data displays possible.</entry>
387 <entry><xref linkend="special-retrieval"/></entry>
394 <section id="features-sort-rank">
395 <title>&zebra; Sorting and Ranking</title>
397 <table id="table-features-sort-rank" frame="top">
398 <title>&zebra; sorting and ranking</title>
400 <colspec colwidth="1*" colname="feature"/>
401 <colspec colwidth="1*" colname="availability"/>
402 <colspec colwidth="3*" colname="notes"/>
403 <colspec colwidth="2*" colname="references"/>
406 <entry>Feature</entry>
407 <entry>Availability</entry>
409 <entry>Reference</entry>
415 <entry>numeric, lexicographic</entry>
416 <entry>Sorting on the basis of alpha-numeric and numeric data
417 is supported. Alphanumeric sorts can be configured for
418 different data encodings and locales for European languages.</entry>
419 <entry><xref linkend="administration-ranking-sorting"/> and
420 <xref linkend="querymodel-zebra-attr-sorting"/></entry>
423 <entry>Combined sorting</entry>
425 <entry>Sorting on the basis of combined sorts  e.g. combinations of
426 ascending/descending sorts of lexicographical/numeric/date field data
428 <entry><xref linkend="administration-ranking-sorting"/></entry>
431 <entry>Relevance ranking</entry>
432 <entry>TF-IDF like</entry>
433 <entry>Relevance-ranking of free-text queries is supported
434 using a TF-IDF like algorithm.</entry>
435 <entry><xref linkend="administration-ranking-dynamic"/></entry>
438 <entry>Static pre-ranking</entry>
440 <entry>Enables pre-index time ranking of documents where hit
441 lists are ordered first by ascending static rank, then by
442 ascending document ID.</entry>
443 <entry><xref linkend="administration-ranking-static"/></entry>
451 <section id="features-updates">
452 <title>&zebra; Live Updates</title>
455 <table id="table-features-updates" frame="top">
456 <title>&zebra; live updates</title>
458 <colspec colwidth="1*" colname="feature"/>
459 <colspec colwidth="1*" colname="availability"/>
460 <colspec colwidth="3*" colname="notes"/>
461 <colspec colwidth="2*" colname="references"/>
464 <entry>Feature</entry>
465 <entry>Availability</entry>
467 <entry>Reference</entry>
472 <entry>Incremental and batch updates</entry>
474 <entry>It is possible to schedule record inserts/updates/deletes in any
475 quantity, from single individual handled records to batch updates
476 in strikes of any size, as well as total re-indexing of all records
477 from file system. </entry>
478 <entry><xref linkend="zebraidx"/></entry>
481 <entry>Remote updates</entry>
482 <entry>&z3950; extended services</entry>
483 <entry>Updates can be performed from remote locations using the
484 &z3950; extended services. Access to extended services can be
485 login-password protected.</entry>
486 <entry><xref linkend="administration-extended-services"/> and
487 <xref linkend="zebra-cfg"/></entry>
490 <entry>Live updates</entry>
491 <entry>transaction based</entry>
492 <entry> Data updates are transaction based and can be performed
493 on running &zebra; systems. Full searchability is preserved
494 during life data update due to use of shadow disk areas for
495 update operations. Multiple update transactions at the same
496 time are lined up, to be performed one after each other. Data
497 integrity is preserved.</entry>
498 <entry><xref linkend="shadow-registers"/></entry>
505 <section id="features-protocol">
506 <title>&zebra; Networked Protocols</title>
508 <table id="table-features-protocol" frame="top">
509 <title>&zebra; networked protocols</title>
511 <colspec colwidth="1*" colname="feature"/>
512 <colspec colwidth="1*" colname="availability"/>
513 <colspec colwidth="3*" colname="notes"/>
514 <colspec colwidth="2*" colname="references"/>
517 <entry>Feature</entry>
518 <entry>Availability</entry>
520 <entry>Reference</entry>
525 <entry>Fundamental operations</entry>
526 <entry>&z3950;/&sru; <literal>explain</literal>,
527 <literal>search</literal>, <literal>scan</literal>, and
528 <literal>update</literal></entry>
530 <entry><xref linkend="querymodel-operation-types"/></entry>
533 <entry>&z3950; protocol support</entry>
535 <entry> Protocol facilities supported are:
536 <literal>init</literal>, <literal>search</literal>,
537 <literal>present</literal> (retrieval),
538 Segmentation (support for very large records),
539 <literal>delete</literal>, <literal>scan</literal>
540 (index browsing), <literal>sort</literal>,
541 <literal>close</literal> and support for the <literal>update</literal>
542 Extended Service to add or replace an existing &xml;
543 record. Piggy-backed presents are honored in the search
544 request. Named result sets are supported.</entry>
545 <entry><xref linkend="protocol-support"/></entry>
548 <entry>Web Service support</entry>
549 <entry>&sru_gps;</entry>
550 <entry> The protocol operations <literal>explain</literal>,
551 <literal>searchRetrieve</literal> and <literal>scan</literal>
552 are supported. <ulink url="&url.cql;">&cql;</ulink> to internal
554 conversion is supported. Extended RPN queries
555 for search/retrieve and scan are supported.</entry>
556 <entry><xref linkend="zebrasrv-sru-support"/></entry>
563 <section id="features-scalability">
564 <title>&zebra; Data Size and Scalability</title>
566 <table id="table-features-scalability" frame="top">
567 <title>&zebra; data size and scalability</title>
569 <colspec colwidth="1*" colname="feature"/>
570 <colspec colwidth="1*" colname="availability"/>
571 <colspec colwidth="3*" colname="notes"/>
572 <colspec colwidth="2*" colname="references"/>
575 <entry>Feature</entry>
576 <entry>Availability</entry>
578 <entry>Reference</entry>
583 <entry>No of records</entry>
584 <entry>40-60 million</entry>
589 <entry>Data size</entry>
590 <entry>100 GB of record data</entry>
591 <entry>&zebra; based applications have successfully indexed up
592 to 100 GB of record data</entry>
596 <entry>Scale out</entry>
597 <entry>multiple discs</entry>
602 <entry>Performance</entry>
603 <entry><literal>O(n * log N)</literal></entry>
604 <entry> &zebra; query speed and performance is affected roughly by
605 <literal>O(log N)</literal>,
606 where <literal>N</literal> is the total database size, and by
607 <literal>O(n)</literal>, where <literal>n</literal> is the
608 specific query hit set size.</entry>
612 <entry>Average search times</entry>
614 <entry> Even on very large size databases hit rates of 20 queries per
615 seconds with average query answering time of 1 second are possible,
616 provided that the boolean queries are constructed sufficiently
617 precise to result in hit sets of the order of 1000 to 5.000
622 <entry>Large databases</entry>
623 <entry>64 bit file pointers</entry>
624 <entry>64 file pointers assure that register files can extend
625 the 2 GB limit. Logical files can be
626 automatically partitioned over multiple disks, thus allowing for
627 large databases.</entry>
635 <section id="features-platforms">
636 <title>&zebra; Supported Platforms</title>
638 <table id="table-features-platforms" frame="top">
639 <title>&zebra; supported platforms</title>
641 <colspec colwidth="1*" colname="feature"/>
642 <colspec colwidth="1*" colname="availability"/>
643 <colspec colwidth="3*" colname="notes"/>
644 <colspec colwidth="2*" colname="references"/>
647 <entry>Feature</entry>
648 <entry>Availability</entry>
650 <entry>Reference</entry>
657 <entry>GNU Linux (32 and 64bit), journaling Reiser or (better)
659 on disks. NFS file systems are not supported.
660 GNU/Debian Linux packages are available</entry>
661 <entry><xref linkend="installation-debian"/></entry>
665 <entry>tar-ball</entry>
666 <entry>&zebra; is written in portable C, so it runs on most
668 Usual tar-ball install possible on many major Unix systems</entry>
669 <entry><xref linkend="installation-unix"/></entry>
672 <entry>Windows</entry>
673 <entry>NT/2000/2003/XP</entry>
674 <entry>&zebra; runs as well on Windows (NT/2000/2003/XP).
675 Windows installer packages available</entry>
676 <entry><xref linkend="installation-win32"/></entry>
686 <section id="introduction-apps">
687 <title>References and &zebra; based Applications</title>
689 &zebra; has been deployed in numerous applications, in both the
690 academic and commercial worlds, in application domains as diverse
691 as bibliographic catalogues, Geo-spatial information, structured
692 vocabulary browsing, government information locators, civic
693 information systems, environmental observations, museum information
697 Notable applications include the following:
701 <section id="koha-ils">
702 <title>Koha free open-source ILS</title>
704 <ulink url="http://www.koha.org/">Koha</ulink> is a full-featured
705 open-source ILS, initially developed in
706 New Zealand by Katipo Communications Ltd, and first deployed in
707 January of 2000 for Horowhenua Library Trust. It is currently
708 maintained by a team of software providers and library technology
709 staff from around the globe.
712 <ulink url="http://liblime.com/">LibLime</ulink>,
713 a company that is marketing and supporting Koha, adds in
714 the new release of Koha 3.0 the &zebra;
715 database server to drive its bibliographic database.
718 In early 2005, the Koha project development team began looking at
719 ways to improve &marc; support and overcome scalability limitations
720 in the Koha 2.x series. After extensive evaluations of the best
721 of the Open Source textual database engines - including MySQL
722 full-text searching, PostgreSQL, Lucene and Plucene - the team
726 "&zebra; completely eliminates scalability limitations, because it
727 can support tens of millions of records." explained Joshua
728 Ferraro, LibLime's Technology President and Koha's Project
729 Release Manager. "Our performance tests showed search results in
730 under a second for databases with over 5 million records on a
731 modest i386 900Mhz test server."
734 "&zebra; also includes support for true boolean search expressions
735 and relevance-ranked free-text queries, both of which the Koha
736 2.x series lack. &zebra; also supports incremental and safe
737 database updates, which allow on-the-fly record
738 management. Finally, since &zebra; has at its heart the &z3950;
739 protocol, it greatly improves Koha's support for that critical
743 Although the bibliographic database will be moved to &zebra;, Koha
744 3.0 will continue to use a relational SQL-based database design
745 for the 'factual' database. "Relational database managers have
746 their strengths, in spite of their inability to handle large
747 numbers of bibliographic records efficiently," summed up Ferraro,
748 "We're taking the best from both worlds in our redesigned Koha
752 See also LibLime's newsletter article
753 <ulink url="http://www.liblime.com/newsletter/2006/01/features/koha-earns-its-stripes/">
754 Koha Earns its Stripes</ulink>.
758 <section id="emilda-ils">
759 <title>Emilda open source ILS</title>
761 <ulink url="http://www.emilda.org/">Emilda</ulink>
762 is a complete Integrated Library System, released under the
763 GNU General Public License. It has a
764 full featured Web-OPAC, allowing comprehensive system management
765 from virtually any computer with an Internet connection, has
766 template based layout allowing anyone to alter the visual
767 appearance of Emilda, and is
768 &xml; based language for fast and easy portability to virtually any
770 Currently, Emilda is used at three schools in Espoo, Finland.
773 As a surplus, 100% &marc; compatibility has been achieved using the
774 &zebra; Server from Index Data as backend server.
778 <section id="reindex-ils">
779 <title>ReIndex.Net web based ILS</title>
781 <ulink url="http://www.reindex.net/index.php?lang=en">Reindex.net</ulink>
782 is a netbased library service offering all
783 traditional functions on a very high level plus many new
784 services. Reindex.net is a comprehensive and powerful WEB system
785 based on standards such as &xml; and &z3950;.
786 updates. Reindex supports &marc21;, dan&marc; eller Dublin Core with
790 Reindex.net runs on GNU/Debian Linux with &zebra; and Simpleserver
792 Data for bibliographic data. The relational database system
793 Sybase 9 &xml; is used for
795 Internally &marcxml; is used for bibliographical records. Update
796 utilizes &z3950; extended services.
800 <section id="dads-article-database">
801 <title>DADS - the DTV Article Database
804 DADS is a huge database of more than ten million records, totalling
805 over ten gigabytes of data. The records are metadata about academic
806 journal articles, primarily scientific; about 10% of these
807 metadata records link to the full text of the articles they
808 describe, a body of about a terabyte of information (although the
809 full text is not indexed.)
812 It allows students and researchers at DTU (Danmarks Tekniske
813 Universitet, the Technical College of Denmark) to find and order
814 articles from multiple databases in a single query. The database
815 contains literature on all engineering subjects. It's available
816 on-line through a web gateway, though currently only to registered
820 More information can be found at
821 <ulink url="http://www.dtv.dk/"/> and
822 <ulink url="http://dads.dtv.dk"/>
826 <section id="infonet-eprints">
827 <title>Infonet Eprints</title>
829 The InfoNet Eprints service from the
830 <ulink url="http://www.dtv.dk/">
831 Technical Knowledge Center of Denmark</ulink>
832 provides access to documents stored in
833 eprint/preprint servers and institutional research archives around
834 the world. The service is based on Open Archives Initiative metadata
835 harvesting of selected scientific archives around the world. These
836 open archives offer free and unrestricted access to their contents.
839 Infonet Eprints currently holds 1.4 million records from 16 archives.
840 The online search facility is found at
841 <ulink url="http://preprints.cvt.dk"/>.
845 <section id="alvis-project">
848 The <ulink url="http://www.alvis.info/alvis/">Alvis</ulink> EU
849 project run under the 6th Framework (IST-1-002068-STP)
850 is building a semantic-based peer-to-peer search engine. A
851 consortium of eleven partners from six different European
852 Community countries plus Switzerland and China contribute
853 with expertise in a broad range of specialties including network
854 topologies, routing algorithms, linguistic analysis and
858 The &zebra; information retrieval indexing machine is used inside
859 the Alvis framework to
860 manage huge collections of natural language processed and
861 enhanced &xml; data, coming from a topic relevant web crawl.
862 In this application, &zebra; swallows and manages 37GB of &xml; data
863 in about 4 hours, resulting in search times of fractions of
870 <title>ULS (Union List of Serials)</title>
873 has created a union catalogue for the periodicals of the
874 twenty-one constituent libraries of the University of London and
875 the University of Westminster
876 (<ulink url="http://www.m25lib.ac.uk/ULS/"/>).
877 They have achieved this using an
878 unusual architecture, which they describe as a
879 ``non-distributed virtual union catalogue''.
882 The member libraries send in data files representing their
883 periodicals, including both brief bibliographic data and summary
884 holdings. Then 21 individual &z3950; targets are created, each
885 using &zebra;, and all mounted on the single hardware server.
886 The live service provides a web gateway allowing &z3950; searching
887 of all of the targets or a selection of them. &zebra;'s small
888 footprint allows a relatively modest system to comfortably host
892 More information can be found at
893 <ulink url="http://www.m25lib.ac.uk/ULS/"/>
898 <title>NLI-&z3950; - a Natural Language Interface for Libraries</title>
900 Fernuniversität Hagen in Germany have developed a natural
901 language interface for access to library databases.
903 url="http://ki212.fernuni-hagen.de/nli/NLIintro.html"/> -->
904 In order to evaluate this interface for recall and precision, they
905 chose &zebra; as the basis for retrieval effectiveness. The &zebra;
906 server contains a copy of the GIRT database, consisting of more
907 than 76000 records in &sgml; format (bibliographic records from
908 social science), which are mapped to &marc; for presentation.
911 (GIRT is the German Indexing and Retrieval Testdatabase. It is a
912 standard German-language test database for intelligent indexing
913 and retrieval systems. See
914 <ulink url="http://www.gesis.org/forschung/informationstechnologie/clef-delos.htm"/>)
917 Evaluation will take place as part of the TREC/CLEF campaign 2003
918 <ulink url="http://clef.iei.pi.cnr.it"/>.
919 <!-- or <ulink url="http://www4.eurospider.ch/CLEF/"/> -->
922 For more information, contact Johannes Leveling
923 <email>Johannes.Leveling@FernUni-Hagen.De</email>
927 <section id="various-web-indexes">
928 <title>Various web indexes</title>
930 &zebra; has been used by a variety of institutions to construct
931 indexes of large web sites, typically in the region of tens of
932 millions of pages. In this role, it functions somewhat similarly
933 to the engine of google or altavista, but for a selected intranet
934 or a subset of the whole Web.
937 For example, Liverpool University's web-search facility (see on
939 <ulink url="http://www.liv.ac.uk/"/>
940 and many sub-pages) works by relevance-searching a &zebra; database
941 which is populated by the Harvest-NG web-crawling software.
944 For more information on Liverpool university's intranet search
945 architecture, contact John Gilbertson
946 <email>jgilbert@liverpool.ac.uk</email>
950 has recently modified the Harvest web indexer to use &zebra; as
951 its native repository engine. His comments on the switch over
952 from the old engine are revealing:
955 The first results after some testing with &zebra; are very
956 promising. The tests were done with around 220,000 SOIF files,
957 which occupies 1.6GB of disk space.
960 Building the index from scratch takes around one hour with &zebra;
961 where [old-engine] needs around five hours. While [old-engine]
962 blocks search requests when updating its index, &zebra; can still
963 answer search requests.
965 &zebra; supports incremental indexing which will speed up indexing
969 While the search time of [old-engine] varies from some seconds
970 to some minutes depending how expensive the query is, &zebra;
971 usually takes around one to three seconds, even for expensive
974 &zebra; can search more than 100 times faster than [old-engine]
975 and can process multiple search requests simultaneously
978 I am very happy to see such nice software available under GPL.
986 <section id="introduction-support">
987 <title>Support</title>
989 You can get support for &zebra; from at least three sources.
992 First, there's the &zebra; web site at
993 <ulink url="&url.idzebra;"/>,
994 which always has the most recent version available for download.
995 If you have a problem with &zebra;, the first thing to do is see
996 whether it's fixed in the current release.
999 Second, there's the &zebra; mailing list. Its home page at
1000 <ulink url="&url.idzebra.mailinglist;"/>
1001 includes a complete archive of all messages that have ever been
1002 posted on the list. The &zebra; mailing list is used both for
1003 announcements from the authors (new
1004 releases, bug fixes, etc.) and general discussion. You are welcome
1005 to seek support there. Join by filling the form on the list home page.
1008 Third, it's possible to buy a commercial support contract, with
1009 well defined service levels and response times, from Index Data.
1011 <ulink url="&url.indexdata.support;"/>
1016 <!-- Keep this comment at the end of the file
1021 sgml-minimize-attributes:nil
1022 sgml-always-quote-attributes:t
1025 sgml-parent-document: "zebra.xml"
1026 sgml-local-catalogs: nil
1027 sgml-namecase-general:t