Historic Pittsburgh, digitalis gyujtemeny

[Király Péter]

Az ev site-ja?

A http://digital.library.pitt.edu/pittsburgh/ cimen egy igen erdekes
gyujtemeny talalhato "Torteneti Pittsburgh" cimen. Az alkotok tobb mint 60
muvet dolgoztak fel a kovetkezo modon: a lapokat beszkenneltek es az olvaso
kulonfele felbontasokban (100% [gif kep, kb. 85-90 KB], 75% [60 KB], 50% [20
KB], 25% [12 KB] - megjegyzendo, hogy ha nem is tisztan de meg ez a
felbontas is olvashato, legalabbis megallapithato, hogy az oldalon relevans
informaciok szerepelnek-e) ezeket latja bongeszes kozben. Navigacios
lehetoseg: elore egyet, hatra egyet, ill. tetszoleges oldalra ugras. Ilyet
mar lattunk jonehanyat, ez meg onmagaban nem kulonleges. Azonban az
oldalakat karakterfelismero programmal fel is ismertettek es nehany igen jo
visszakeresesi lehetoseget biztositottak

1) egyszeru kereses: pontos kifejezesre vagy szotoredekre (ha pontos
kifejezest keresunk a kif. utan tegyunk egy szokozt)
2) boole algebras kereses: u.az mint az elozo, de itt tobb szot es ezek
kapcsolatat (es, vagy, nem) is megadhatjuk, valamint a kapcsolat kornyezetet
(oldal, fejezet, mu) (pl. keressuk azt az oldalt, ahol szerepel a "roman" es
a "magyar" szo is, vagy azt a muvet amiben szerepel "Kossuth" de "Lajos" nem
stb.)
3) "kozelisegi" kereses: ket vagy harom szo (vagy szotoredek) kozott fennall
a kovetkezo kapcsolat  a az elso szo "mellett" vagy "utan" vagy "mellette
nem" vagy "utanna nem" all 40/80/120 karakternyi tavolsagra a kovetkezo szo.
A mellett es az utan kozott az a kulonbseg, hogy a mellett nem határoz meg
sorrendiseget, lehet az elso szo elott is es utanna is. (pl. Kossuth
"mellett" Louis kihozza a Louis Kossuthot is, mig az "utan" csak a "Kossuth,
Louis"-t)
4) Bibliografiai kereses: A kovetkezo mezokben lehet keresni: cim, szerzo,
targy, kiadasi adatok (pl. Pittsburgh, 1844). A mezok kozott logikai
kapcsolatot lehet epiteni (ld. 2) kereses)

megj. 1.: a fenti kifejezeseket termeszetesen angolul lehet az eredeti lapon
alkalmazni.
megj. 2.: Pittsburghban talalhato - talan a mai napig, bar errol nincs
pontos informaciom - az Egyesult Allamok egyik legnagyobb magyar kozossege.
Ennek torteneterol a site-on rengeteg anyagot talalhatunk.

A http://digital.library.pitt.edu/pittsburgh/vendors.html cimen a keszites
meneterol talalhatunk informaciokat:
szovegszkenneles: 600 dpi, Group IV. TIFF formatumban
terkepszkenneles: 300 dpi, 24 bites szinmelysegu tomoritetlen TIFF
szovegfelismeres: Prime Recognition szoftver
korrektura: egeszen minimalis melysegig
visszakereses: mind a szovegek, mind az EAD segedletek OpenText SGML-kereso,
a terkepek MySQL-lel
kepkonvertalas: a Lizardteh MrSID programja (ez szerver-kliens
alkalmazasokban "ropteben" kepes kulonfele kepmereteket eloallitani a sajat
maguk kifejlesztette formatumrol)
CGI (nem emlitettem, hogy az egesz site dinamikus HTML lapokból épül fel): a
Michigani Egyetem kész SGML-lekerdezo szrkiptjeit alkalmaztak a helyi
viszonyokra.

A sitehoz egy igen alapos Sugo is tartozik, amely feloleli a kereses, a
nyomtatas, az idezes es a navigalas kerdeseit, tovabba innen erhetoek el a
pittsburghi kozos leveltari katalogus rekordjai is (a Pittburghi Egyetemi
Konyvtar es a Nyugat-Pennsylvaniai Torteneti Tarsulat gyujtemenyeinek
segedletei). A leirasok EAD (Kodolt Leveltari Leiras, mar tobbszor volt szo
rola a listan) formaban keszultek. A segedletek kovetkezokeppen lehet
keresni: "barhol", "gyujtemenyi szintu leirasban", "irat szintu leírásban"
(contents list), "targyszavak kozott" (controlled access terms).

Egyszoval, ha engem kerdeznenek meg, en biztosan nekik adnam az "ev
torteneti honlapja" vagy az "ev elektronikus gyujtemenye" cimet.

megj. 3.: kulon orom, hogy a gyujtemeny nem valt ilyen-olyan cegek
reklamjava. Az egyetlen neven nevezett tamogato a Hillman Konyvtari
Alapitvany volt.

(kirunews)

----- Original Message -----
From: Doris Hayashikawa <dorish+ at pitt.edu>
To: <history-digitisation at mailbase.ac.uk>
Sent: Wednesday, November 10, 1999 11:04 PM
Subject: Historic Pittsburgh, a Digital Collection


>List members:  We would appreciate feedback on our work. Much of it is
>still in development.  We also have plans for user evaluation studies
>based on our digital collections in the near future.
>_________________
>The Digital Research Library, a division of the University Library
>System, University of Pittsburgh, is pleased to announce the
>availability of the collection, Historic Pittsburgh, at this URL:
>http://digital.library.pitt.edu/pittsburgh/
>
>Digital editions of sixty-two volumes on the history of Western
>Pennsylvania are currently available. These materials cover the growth
>and development of Pittsburgh and the surrounding area from the period
>of exploration and settlement to the period of industrial revolution and
>modernization. Items selected for digitization are from the collections
>of the University Library System and from the Library of the Historical
>Society of Western Pennsylvania, a partner institution.
>
>The contents of the digital editions are fully searchable for ease in
>finding pages on which specific names, places, events, etc. occur.
>Additional navigation through the books is made simple via hyperlinked
>tables of contents which allow the user go directly to individual
>chapters and illustrations within a book.  When complete, the collection
>will contain over four hundred titles.
>
>In addition, 19th and early 20th century real estate plat maps
>originally published by the G.M. Hopkins Company have been converted to
>digital images and may be viewed using an ordinary Web browser. MrSid,
>compression software available from LizardTech, is the underlying
>technology. Eighty-five maps are currently on-line out of a total of six
>hundred that will eventually be made available. Ultimately this
>collection will reflect the changing urban landscape of the greater
>Pittsburgh area from the early 1870s to the late 1930s. The plat maps
>show lot and block numbers, dimensions, street widths, and names of
>property owners as well as churches, cemeteries, mills, schools, roads,
>railroads, lakes, ponds, rivers, and streams.  Currently, extensive
>metadata are being collected from each map for use in a retrieval
>database which will be developed in the next month or two.
>
>The On-line Archival Finding Aids project provides convenient Web access
>to detailed descriptions of archival collections held by the University
>Library System, University of Pittsburgh, and the Historical Society of
>Western Pennsylvania.  This collection of guides to archival collections
>will enable students and scholars to more easily access important
>primary source materials held by these two repositories and to identify
>items relevant to their research.  The preliminary collection of
>sixty-three finding aids will be augmented monthly as more finding aids
>are encoded for Web accessibility.  The EAD standard is being used for
>this project.  A Web based input form to facilitate the writing of
>finding aids directly into EAD encoded form by staff without extensive
>training is currently under development.
>
>The full-text collection and the EAD encoded finding aids are made
>searchable through use of the PAT50 search engine of OpenText.
>Middleware used in our implementation was adapted from work done at the
>Digital Production Services, University of Michgan. It is made available
>to us through membership in their SGML Server program. Searchable text
>for the full-text collection is generated in batch mode from TIFF images
>using five OCR engines of PrimeRecognition.  Staff of the DRL edit OCR
>output minimally. Pages describing these technologies and our processes
>are available through the Historic Pittsburgh Web site.
>
>
>--
>Doris Hayashikawa, Coordinator
>Digital Research Library, University Library System
>University of Pittsburgh
>G20R Hillman Library
>Pittsburgh, PA 15260
>Phone:  (412) 648-7765
>
>