1 <chapter id="introduction">
2 <!-- $Id: introduction.xml,v 1.34 2006-06-30 13:58:07 marc Exp $ -->
3 <title>Introduction</title>
6 <title>Overview</title>
9 <ulink url="http://indexdata.dk/zebra/">Zebra</ulink>
10 is a high-performance, general-purpose structured text
11 indexing and retrieval engine. It reads records in a
12 variety of input formats (eg. email, XML, MARC) and provides access
13 to them through a powerful combination of boolean search
14 expressions and relevance-ranked free-text queries.
18 Zebra supports large databases (tens of millions of records,
19 tens of gigabytes of data). It allows safe, incremental
20 database updates on live systems. Because Zebra supports
21 the industry-standard information retrieval protocol, Z39.50,
22 you can search Zebra databases using an enormous variety of
23 programs and toolkits, both commercial and free, which understand
24 this protocol. Application libraries are available to allow
25 bespoke clients to be written in Perl, C, C++, Java, Tcl, Visual
26 Basic, Python, PHP and more - see
27 <ulink url="http://zoom.z3950.org/">the ZOOM web site</ulink>
28 for more information on some of these client toolkits.
32 This document is an introduction to the Zebra system. It explains
33 how to compile the software, how to prepare your first database,
34 and how to configure the server to give you the
35 functionality that you need.
40 <title>Features</title>
43 This is an overview of some of Zebra's most important features:
51 Very large databases: logical files can be
52 automatically partitioned over multiple disks.
58 Arbitrarily complex records. The internal data format
59 is a structured format conceptually similar to XML or GRS-1,
60 which allows lists, nested structured data elements and
61 variant forms of data.
67 Robust updating - records can be added and deleted ``on the fly''
68 without rebuilding the index from scratch.
69 Records can be safely updated even while users are accessing
71 The update procedure is tolerant to crashes or hard interrupts
72 during database updating - data can be reconstructed following
79 Configurable to understand many input formats.
80 A system of input filters driven by
81 regular expressions allows most ASCII-based
82 data formats to be easily processed.
83 SGML, XML, ISO2709 (MARC), and raw text are also
90 Searching supports a powerful combination of boolean queries as
91 well as relevance-ranking (free-text) queries. Truncation,
92 masking, full regular expression matching and "approximate
93 matching" (eg. spelling mistakes) are all handled.
99 Index-only databases: data can be, and usually is, imported
100 into Zebra's own storage, but Zebra can also refer to
101 external files, building and maintaining indexes of "live"
108 Zebra is written in portable C, so it runs on most Unix-like systems
109 as well as Windows NT. A binary distribution for Windows NT is
111 <ulink url="http://ftp.indexdata.dk/pub/zebra/win32/"/>,
112 and pre-built packages are available for
116 <ulink url="http://ftp.indexdata.dk/pub/zebra/RedHat7.X/"/>
117 and Debian packages at
119 <literal>GNU/Debian Linux</literal> at
120 <ulink url="http://ftp.indexdata.dk/pub/zebra/debian/"/>.
129 <ulink url="&url.z39.50;">Z39.50</ulink> protocol support:
136 Protocol facilities: Init, Search, Present (retrieval),
137 Segmentation (support for very large records), Delete, Scan
138 (index browsing), Sort, Close and support for the ``update''
139 Extended Service to add or replace an existing XML record.
145 Piggy-backed presents are honored in the search request - that
146 is, a subset of the found records can be returned directly with
147 a search response, enabling search and retrieval to happen in a
154 Named result sets are supported.
160 Easily configured to support different application profiles, with
161 tables for attribute sets, tag sets, and abstract syntaxes.
162 Additional tables control facilities such as element mappings to
163 different schema (eg., GILS-to-USMARC).
169 Complex composition specifications using Espec-1 (partial support).
170 Element sets are defined using the Espec-1 capability,
171 and are specified in configuration files as simple element
172 requests (and, optionally, variant requests).
178 Multiple record syntaxes
179 for data retrieval: GRS-1, SUTRS,
180 XML, ISO2709 (MARC), etc. Records can be mapped between record syntaxes
181 and schemas on the fly.
191 <ulink url="&url.sru;">SRU</ulink> Web Service support:
197 The protocol operations <literal>explain</literal>,
198 <literal>searchRetrieve</literal> and <literal>scan</literal>
204 <ulink url="&url.cql;">CQL</ulink> to internal query model RPN
205 conversion is supported.
210 Multiple XML record formats
211 for data retrieval are supported, modelled over the GRS-1, SUTRS,
212 MARC record formats. Records can be mapped between record
213 schemas on the fly. Arbitrarily complex XSLT transformations
214 can be applied during record retrieval if one uses the
215 <literal>alvis</literal> filter module.
220 Additional PQF query syntax for
221 <literal>searchRetrieve</literal>
222 and <literal>scan</literal> operations is supported.
233 <sect1 id="introduction-apps">
234 <title>References and Zebra based Applications</title>
236 Zebra has been deployed in numerous applications, in both the
237 academic and commercial worlds, in application domains as diverse
238 as bibliographic catalogues, geospatial information, structured
239 vocabulary browsing, government information locators, civic
240 information systems, environmental observations, museum information
244 Notable applications include the following:
249 <title>Koha free open-source ILS</title>
251 <ulink url="http://www.koha.org/">Koha</ulink> is a full-featured
252 open-source ILS, initially developed in
253 New Zealand by Katipo Communications Ltd, and first deployed in
254 January of 2000 for Horowhenua Library Trust. It is currently
255 maintained by a team of software providers and library technology
256 staff from around the globe.
259 <ulink url="http://liblime.com/">LibLime</ulink>,
260 a company that is marketing and supporting Koha, adds in
261 the new release of Koha 3.0 the Zebra
262 database server to drive its bibliographic database.
265 In early 2005, the Koha project development team began looking at
266 ways to improve MARC support and overcome scalability limitations
267 in the Koha 2.x series. After extensive evaluations of the best
268 of the Open Source textual database engines - including MySQL
269 full-text searching, PostgreSQL, Lucene and Plucene - the team
273 "Zebra completely eliminates scalability limitations, because it
274 can support tens of millions of records." explained Joshua
275 Ferraro, LibLime's Technology President and Koha's Project
276 Release Manager. "Our performance tests showed search results in
277 under a second for databases with over 5 million records on a
278 modest i386 900Mhz test server."
281 "Zebra also includes support for true boolean search expressions
282 and relevance-ranked free-text queries, both of which the Koha
283 2.x series lack. Zebra also supports incremental and safe
284 database updates, which allow on-the-fly record
285 management. Finally, since Zebra has at its heart the Z39.50
286 protocol, it greatly improves Koha's support for that critical
290 Although the bibliographic database will be moved to Zebra, Koha
291 3.0 will continue to use a relational SQL-based database design
292 for the 'factual' database. "Relational database managers have
293 their strengths, in spite of their inability to handle large
294 numbers of bibliographic records efficiently," summed up Ferraro,
295 "We're taking the best from both worlds in our redesigned Koha
301 <title>Emilda open source ILS</title>
303 <ulink url="http://www.emilda.org/">Emilda</ulink>
304 is a complete Integrated Library System, released under the
305 GNU General Public License. It has a
306 full featured Web-OPAC, allowing comprehensive system management
307 from virtually any computer with an Internet connection, has
308 template based layout allowing anyone to alter the visual
309 appearance of Emilda, and is
310 XML based language for fast and easy portability to virtually any
312 Currently, Emilda is used at three schools in Espoo, Finland.
315 As a surplus, 100% MARC compatibility has been achieved using the
316 Zebra Server from Index Data as backend server.
321 <title>ReIndex.Net web based ILS</title>
323 <ulink url="http://www.reindex.net/index.php?lang=en">Reindex.net</ulink>
324 is a netbased library service offering all
325 traditional functions on a very high level plus many new
326 services. Reindex.net is a comprehensive and powerful WEB system
327 based on standards such as XML and Z39.50.
328 updates. Reindex supports MARC21, danMARC eller Dublin Core with
332 Reindex.net runs on GNU/Debian Linux with Zebra and Simpleserver
334 Data for bibliographic data. The reational database system
335 Sybase 9 XML is used for
337 Internally MARCXML is used for bibliographical records. Update
338 utilizes Z39.50 extended services.
344 <title>DADS - the DTV Article Database Service</title>
346 DADS is a huge database of more than ten million records, totalling
347 over ten gigabytes of data. The records are metadata about academic
348 journal articles, primarily scientific; about 10% of these
349 metadata records link to the full text of the articles they
350 describe, a body of about a terabyte of information (although the
351 full text is not indexed.)
354 It allows students and researchers at DTU (Danmarks Tekniske
355 Universitet, the Technical College of Denmark) to find and order
356 articles from multiple databases in a single query. The database
357 contains literature on all engineering subjects. It's available
358 on-line through a web gateway, though currently only to registered
362 More information can be found at
363 <ulink url="http://www.dtv.dk/"/> and
364 <ulink url="http://dads.dtv.dk"/>
369 <title>Infonet Eprints</title>
371 The InfoNet Eprints service from the
372 <ulink url="http://www.dtv.dk/">
373 Technical Knowledge Center of Denmark</ulink>
374 provides access to documents stored in
375 eprint/preprint servers and institutional research archives around
376 the world. The service is based on Open Archives Initiative metadata
377 harvesting of selected scientific archives around the world. These
378 open archives offer free and unrestricted access to their contents.
381 Infonet Eprints currently holds 1.4 million records from 16 archives.
382 The online search facility is found at
383 <ulink url="http://preprints.cvt.dk"/>.
390 The <ulink url="http://www.alvis.info/alvis/">Alvis</ulink> EU
391 project run under the 6th Framework (IST-1-002068-STP)
392 is building a semantic-based peer-to-peer search engine. A
393 consortium of eleven partners from six different European
394 Community countries plus Switzerland and China contribute
395 expertise in a broad range of specialties including network
396 topologies, routing algorithms, linguistic analysis and
400 The Zebra information retrieval indexing machine is used inside
401 the Alvis framework to
402 manage huge collections of natural language processed and
403 enhanced XML data, coming from a topic relevant web crawl.
404 In this application, Zebra swallows and manages 37GB of XML data
405 in about 4 hours, resulting in search timese of fraction of
412 <title>ULS (Union List of Serials)</title>
415 has created a union catalogue for the periodicals of the
416 twenty-one constituent libraries of the University of London and
417 the University of Westminster
418 (<ulink url="http://www.m25lib.ac.uk/ULS/"/>).
419 They have achieved this using an
420 unusual architecture, which they describe as a
421 ``non-distributed virtual union catalogue''.
424 The member libraries send in data files representing their
425 periodicals, including both brief bibliographic data and summary
426 holdings. Then 21 individual Z39.50 targets are created, each
427 using Zebra, and all mounted on the single hardware server.
428 The live service provides a web gateway allowing Z39.50 searching
429 of all of the targets or a selection of them. Zebra's small
430 footprint allows a relatively modest system to comfortably host
434 More information can be found at
435 <ulink url="http://www.m25lib.ac.uk/ULS/"/>
440 <title>NLI-Z39.50 - a Natural Language Interface for Libraries</title>
442 Fernuniversität Hagen in Germany have developed a natural
443 language interface for access to library databases.
445 url="http://ki212.fernuni-hagen.de/nli/NLIintro.html"/> -->
446 In order to evaluate this interface for recall and precision, they
447 chose Zebra as the basis for retrieval effectiveness. The Zebra
448 server contains a copy of the GIRT database, consisting of more
449 than 76000 records in SGML format (bibliographic records from
450 social science), which are mapped to MARC for presentation.
453 (GIRT is the German Indexing and Retrieval Testdatabase. It is a
454 standard German-language test database for intelligent indexing
455 and retrieval systems. See
456 <ulink url="http://www.gesis.org/forschung/informationstechnologie/clef-delos.htm"/>)
459 Evaluation will take place as part of the TREC/CLEF campaign 2003
460 <ulink url="http://clef.iei.pi.cnr.it"/>.
461 <!-- or <ulink url="http://www4.eurospider.ch/CLEF/"/> -->
464 For more information, contact Johannes Leveling
465 <email>Johannes.Leveling@FernUni-Hagen.De</email>
470 <title>Various web indexes</title>
472 Zebra has been used by a variety of institutions to construct
473 indexes of large web sites, typically in the region of tens of
474 millions of pages. In this role, it functions somewhat similarly
475 to the engine of google or altavista, but for a selected intranet
476 or a subset of the whole Web.
479 For example, Liverpool University's web-search facility (see on
481 <ulink url="http://www.liv.ac.uk/"/>
482 and many sub-pages) works by relevance-searching a Zebra database
483 which is populated by the Harvest-NG web-crawling software.
486 For more information on Liverpool university's intranet search
487 architecture, contact John Gilbertson
488 <email>jgilbert@liverpool.ac.uk</email>
492 has recently modified the Harvest web indexer to use Zebra as
493 its native repository engine. His comments on the switch over
494 from the old engine are revealing:
497 The first results after some testing with Zebra are very
498 promising. The tests were done with around 220,000 SOIF files,
499 which occupies 1.6GB of disk space.
502 Building the index from scratch takes around one hour with Zebra
503 where [old-engine] needs around five hours. While [old-engine]
504 blocks search requests when updating its index, Zebra can still
505 answer search requests.
507 Zebra supports incremental indexing which will speed up indexing
511 While the search time of [old-engine] varies from some seconds
512 to some minutes depending how expensive the query is, Zebra
513 usually takes around one to three seconds, even for expensive
516 Zebra can search more than 100 times faster than [old-engine]
517 and can process multiple search requests simultaneously
520 I am very happy to see such nice software available under GPL.
529 <title>Support</title>
531 You can get support for Zebra from at least three sources.
534 First, there's the Zebra web site at
535 <ulink url="http://indexdata.dk/zebra/"/>,
536 which always has the most recent version available for download.
537 If you have a problem with Zebra, the first thing to do is see
538 whether it's fixed in the current release.
541 Second, there's the Zebra mailing list. Its home page at
542 <ulink url="http://lists.indexdata.dk/cgi-bin/mailman/listinfo/zebralist"/>
543 includes a complete archive of all messages that have ever been
544 posted on the list. The Zebra mailing list is used both for
545 announcements from the authors (new
546 releases, bug fixes, etc.) and general discussion. You are welcome
547 to seek support there. Join by filling the form on the list home page.
550 Third, it's possible to buy a commercial support contract, with
551 well defined service levels and response times, from Index Data.
553 <ulink url="http://indexdata.dk/support/"/>
560 <title>Future Directions</title>
563 These are some of the plans that we have for the software in the near
564 and far future, ordered approximately as we expect to work on them.
572 Improved support for XML in search and retrieval. Eventually,
573 the goal is for Zebra to pull double duty as a flexible
574 information retrieval engine and high-performance XML
575 repository. The recent addition of XPath searching is one
576 example of the kind of enhancement we're working on.
579 There is also the experimental <literal>ALVIS XSLT</literal>
580 XML input filter, which unleashes the full power of DOM based
581 XSLT transformations during indexing and record retrieval. Work
582 on this filter has been sponsored by the ALVIS EU project
583 <ulink url="http://www.alvis.info/alvis/"/>. We expect this filter to
584 mature soon, as it is planned to be included in the version 1.4
591 Access to the search engine through SOAP/RPC API to allow the
592 construction of applications without requiring Z39.50 tools.
594 This will shortly be available by means of Index Data's
595 <ulink url="http://www.loc.gov/standards/sru/srw/">SRW</ulink>-to-Z39.50 gateway, currently in beta test.
597 Experimental support of the
598 Search/Retrieve Via URL ( <ulink url="&url.sru;">SRU</ulink>)
599 <ulink url="&url.sru;"/>
600 REST webservice, and the
601 Search/Retrieve Web Service ( <ulink url="http://www.loc.gov/standards/sru/srw/">SRW</ulink>)
602 <ulink url="http://www.loc.gov/standards/sru/srw/"/>
603 SOAP Web Service have recently been added to the YAZ/Zebra
604 combo - including server side Common Query Language (<ulink url="&url.cql;">CQL</ulink>)
605 <ulink url="&url.cql;"/> parsing
606 and configuration. It remains to find a sponsor for further testing,
607 documentation and packaging of this exiting component.
613 Finalisation and documentation of Zebra's C programming
614 API, allowing updates, database management and other functions
615 not readily expressed in Z39.50. We will also consider
616 exposing the API through SOAP.
622 Support for the use of Perl both for access to the Zebra API
623 and for building extension ``plug-ins'' such as input filters.
624 The code for this has been contributed to the source tree by
626 <email>pop@technomat.hu</email>,
627 and is in the process of being integrated and tested.
633 Improved free-text searching. We're first and foremost octet jockeys and
634 we're actively looking for organisations or people who'd like
635 to contribute experience in relevance ranking and text
644 Programmers thrive on user feedback. If you are interested in a
645 facility that you don't see mentioned here, or if there's something
646 you think we could do better, please drop us a mail. Better still,
647 implement it and send us the patches.
650 If you think it's all really neat, you're welcome to drop us a line
651 saying that, too. You can email us on
652 <email>info@indexdata.dk</email>
653 or check the contact info at the end of this manual.
658 <!-- Keep this comment at the end of the file
663 sgml-minimize-attributes:nil
664 sgml-always-quote-attributes:t
667 sgml-parent-document: "zebra.xml"
668 sgml-local-catalogs: nil
669 sgml-namecase-general:t