BibFrame Testing at the University of Washington:

experiences in general and CJK scripts

Charlene ChouUniversity of Washington

atCEAL Committee on Technical Processing meeting

March 26th, 2015

Agenda

• Goals • Training • The core group of BF testing • Testing procedures, observations and outcome• Next steps

Goals• Train cataloging staff to be familiar with linked

data, BibFrame, RDF/XML, Schema, etc.• Review converted data to check for missing data,

wrongly converted data, data with incorrect FRBR attribution, etc. If any problems, report to LC.

• Review converted data for different types of resources, e.g. monographs, serials, AV, e-resources, maps, 3D, scores, still images, etc.

• Review converted data for non-Roman scripts, e.g. CJK, Cyrillic, etc.

A series of training sessions

• BibFrame training session 1-3 + exercises• BibFrame classes: diagram & hierarchical

listing• BibFrame in RDF/XML• Schema.org and RDA/RDF

Core Group

• Head of Cataloging and Metadata Services, Chair

• Principal Cataloger• Metadata Librarian• Lead Catalogers for various types of resources• International Studies– Head of Technical Services of East Asia Library– Slavic Cataloger

Testing procedures, observations and outcome

• Testers select examples/records for the Chair to convert

• Testers verify the converted data against its original MARC record, and have group discussions.

• The Chair reports anything wrong to LC for correction.

• The Chair maintains a BibFrame Issues List with ticket numbers on GitHub (e.g. #XX) as well as an indication whether each issue was submitted by UW or another institution.

UW BIBFRAME Issues List: some examples (94 comments from UW)

• #83 (other): allow the text of nonstandard relationship designators to be recorded• #105 (other): what is the plan for holdings records?• #129 (UW): workTitle as 245 or 130 a, n, p together as one string; when from 240 include $l and probably other

subfields• #144 (UW): Computer Files assigned wrong content-type subclasses• #148 (UW): serialFirstIssue and serialLastIssue need more generic names• #151 (UW): new algorithm for authorizedAccessPoint in Work• #203 (UW): 6XX ind 4 and 7 should not generate madsrdf:isMemberOfMADSScheme• #206 (UW): bf:contentCategory should be in Work not Instance• #207 (UW): bf:formDesignation should have domain Instance or Work, not Title• #222 (UW): bf:cartographicScale is Work information; 034 subfields need labels• #230 (UW): bf:titleType needs a thesaurus, and changes to the range• #233 (UW): The agent link for an original work (from a translation) has only $a• #235 (UW): bf:label in held item link should have both call numbers when there are two• #238 (UW): $3 needs a big-picture solution• #110 (UW): 042 is not converted• #151 (UW): 046 $o, $p are not converted• #220 (UW): 300 $e should be in a separate instance of bf:extent and not combined with $a• #138 (UW): 520 subfield b is not converted• #166 (UW): 546 in languageNote should be in Work not Instance• #214 (UW): 651 is not converted• #123 (UW): example of 700 $i omitted for related work• #126 (UW): 730s generate related works without an indication of language• #168 (UW): 740 2nd indicator 2 should have relationship hasPart; 740 should not have AAP or creator• #229 (UW): 775 may be instance, not work; should not include imprint in access point• #152 (UW): 776 $i is not converted

BIBFRAME Issues List: related to CJK examples• #119 (UW): 020 $q should be retained for display purposes• #174 (UW): 041 gives non-existent language code for Chinese (zho)• #228 (UW): bf:titleStatement should have parallel non-roman script• #224 (UW): 245 $b non-roman should be in bf:subtitle• #201 (UW): bf:responsibilityStatement should retain brackets; not converted for Maps

format; only parallel script not romanization; not converted for book w/o non-roman script

• #225 (UW): 246 $i with non-roman not given• #178 (UW): 250 repeatability--$3 and parallel scripts• #227 (UW): providerStatement should have parallel non-roman script• #226 (UW): 264 $b with non-roman and multiple publishers duplicates non-roman script• #177 (UW): 505 lacks parallel non-roman script• #116 (UW): 650, @en assigned to foreign language terms• #232 (UW): 650 _7 $2 sk should have non-roman strings with dashes and w/o $2• #157 (UW): 8XX authorized access point should be from 8XX not 490• #162 (UW): language qualifiers on parallel script fields seem wrong (multiple examples)• #173 (UW): non-roman fields missing for related Work entry, e.g. original of translation• #179 (UW): multiparts should not generate multiple instances, for each vol.• #231 (UW): Non-roman forms are not authorized access points

Some CJK examples

• Only in romanized form– <http://example.org/99131426860001452agent14> a bf:Agent ;

bf:label "Ino, Kanori," .– bf:titleStatement "Taiwan ta cha ri ji"

• Wrong language encoding– " 伊能嘉矩 , 1867-1925"@zh-hani ;

– <http://example.org/99131426860001452title6> a bf:Title ; bf:titleValue "Ino Kanori no Taiwan tosa nikki.", " 伊能嘉矩の臺湾踏柤日記 ."@zh-hani .

– bf:providerDate "2014-", "2014-"@zh ;

• Missing $q in 020 ISBN– bf:isbn10 <http://isbn.example.org/986213349X> ;– bf:isbn13 <http://isbn.example.org/9789862133491> ;

Some CJK examples (2)

• Missing $3 in 250 edition for variant editions– bf:edition "Di 1 ban.", "Di 2 ban.", " 第 1 版 ."@zh, " 第 2

版 ."@zh ;• Missing Chinese contents note (505)

– bf:contentsNote "Di 1 juan. Shi xue, shi jia yu shi dai -- Di 2 juan. Zhongguo si xiang chuan tong ji qi xian dai bian qian -- Di 3 juan. Ru jia lun li yu shang ren jing shen -- Di 4 juan. Zhongguo zhi shi ren zhi shi de kao cha -- Di 5 juan. Xian dai xue ren yu xue shu -- Di 6 juan. Min zhu zhi du yu xian dai wen ming -- Di 7-8 juan. Wen hua ping lun yu Zhongguo qing huai (shang, xia). -- Di 9 juan. Li shi ren wu kao bian -- Di 10 juan. Song Ming li xue yu zheng zhi wen hua -- Di 11 juan. Lun xue hui you -- Di 12 juan. Guo xue yu Zhongguo ren wen" ;

Some CJK examples (3)

• Missing $5 WaU for local holdings information– " 本套書限量 500 套 , 此本編號為 357 號 ."@zh

• Missing the romanized form– bf:responsibilityStatement " 蘇軾撰 ; 施元之 , 施宿 , 顧禧

註 ."@zh ;• Missing 246 $i Chinese note

<http://example.org/99161808355801452title26> a bf:Title ;bf:titleType "Title on cover of shou zang wen ji:" ;bf:titleValue "Shao yun er shi, jing dian zai xian", " 少蘊而始 , 經典再現 "@zh .

• Missing 830 field for series– 830 0 Taiwan Resource Center for Chinese Studies at UW-EAL.

$5 WaU

• providerStatement should have parallel non-roman scriptbf:providerStatement "Taibei Shi : Guo jia tu shu guan : Da kuai wen hua, 2012." ;

• 264 $b with non-roman and multiple publishers duplicates non-roman scriptbf:publication [ a bf:Provider ; bf:providerDate "2012", " 二零一二 ."@zh ; bf:providerName [ a bf:Organization ; bf:label "Da kuai wen hua", " 國家圖書館 :"@zh, " 大塊文化 "@zh ] ], [ a bf:Provider ; bf:providerDate "2012", " 二零一二 ."@zh ; bf:providerName [ a bf:Organization ; bf:label "Guo jia tu shu guan ", " 國家圖書館 :"@zh, " 大塊文化 "@zh ] ; bf:providerPlace [ a bf:Place ; bf:label "Taibei Shi ", " 臺北市 :"@zh ] ] ;

Some CJK examples (4)

Next Steps

Research and experiments at UW

Joe Kiegel's presentation at UW on 2/19/2015

RDA/RDF vs. BIBFRAME

RDA/RDF is stronger in

• Series

• Notes

• Technical details of a resource

• Inverse properties (e.g. animator/animatorOf)

BIBFRAME is stronger in

• Administrative metadata

• Identifiers

• Subject headings

• Holdings information

• Support for both transcription (literals) and URIs

Joe Kiegel's presentation at UW on 2/19/2015

• Both RDA/RDF and BIBFRAME are able to represent library information as linked data

• Both schemas originate in libraries and must be adopted widely by non-library users in order to succeed

As Linked Data

Joe Kiegel's presentation at UW on 2/19/2015

• RDA/RDF lacks a number of features of a carrier, e.g. administrative metadata

• BIBFRAME is currently not capable of carrying constrained RDA without significant loss in information about classes

• While neither fully works today, a combination of the two is more successful

As Carriers

Joe Kiegel's presentation at UW on 2/19/2015

• http://faculty.washington.edu/kiegel/ld/RDA-RDF-BIBFRAME-hybrid.pdf

Example of a Hybrid Title

Many thanks to my UW colleagues for such great

teamwork! Special thanks to Joe Kiegel for

his vision and leadership for this initiative!