290 likes | 443 Views
Multilingual, Multi-script Catalog Requirements (An Arcadia Project) ________________________. January 29, 2010. Outline _____________________________________________________. Background about the Arcadia non-Roman script project Introductions Orbis vs. YUFind and systems like YUFind
E N D
Multilingual, Multi-script Catalog Requirements(An Arcadia Project)________________________ January 29, 2010
Outline_____________________________________________________ • Background about the Arcadia non-Roman script project • Introductions • Orbis vs. YUFind and systems like YUFind • Requirements discussion • Wrap-up Jan 2010
Project Goals _____________________________________________________ • Gap analysis of multilingual, multi-script functionality in Lucene-Solr-Solrmarc discovery applications (e.g., YUFind) • Identification of desirable functionality • Collaboration opportunities, community interest • Recommendations with level-of-effort analysis Jan 2010
Orbis vs. Yufind_____________________________________________________ Jan 2010
vs Chinese example: “中日韩经济合作的新起点” N-gram tokens, where N=2: <中日><日韩><韩经><经济><济合><合作><作的> <的新> <新起> <起点>
Background: NR Scripts in Catalog Records_____________________________________________________ Jan 2010
JACKPHY_____________________________________________________ • Japanese • Arabic • Chinese • Korean • Persian • Hebrew • Yiddish Jan 2010
One-to-Many (CJK)_____________________________________________________ Example: “Mao Zedong” 毛泽东Simplified 毛澤東 Traditional 毛沢東Kanji (Modern) Jan 2010
One-to-Many (CJK) _____________________________________________________ “Mao Zedong” in simplified Chinese characters retrieves 527 results Jan 2010
One-to-Many (CJK) _____________________________________________________ The same search in traditional Chinese characters yields154 hits. Also Note paired fields Jan 2010
One-to-Many (Digraphs)_____________________________________________________ The Yiddish word “Virtshaft” is entered here with two separate vavs (i.e., key stroke ‘u’ in Microsoft’s Hebrew IME): U05D5 + U05D5 ווירטשאפט Jan 2010
One-to-Many (Digraphs) _____________________________________________________ N = 49 results Jan 2010
One-to-Many (Digraphs)_____________________________________________________ The same word is this time entered as a double-vav digraph = U05F0 (via MS Hebrew IME key combo right-alt+u) װירטשאפט Jan 2010
One-to-Many (Digraphs)_____________________________________________________ N = 11 results Jan 2010
NR Spelling Suggestions_____________________________________________________ Unhelpful suggestion? Jan 2010
Labels and Facets_____________________________________________________ Should script/language of query determine script/language of facets? Jan 2010
Labels and Facets_____________________________________________________ OR: Sugimoto, Tsutomu, 1927- (11) Takahashi, Mikio, 1935- (11) Noguchi, Takehiko. (8) Watanabe, Shin’ichirō, 1934- (7) Better would be: 杉本つとむ, 1927- (11) 高橋幹夫, 1935- (11) 野口武彦. (8) 渡辺信一郎, 1934- (7) But not both mixed together. Let end user decide? Jan 2010
Labels and Facets_____________________________________________________ • We would like to ask library users the best option for displaying parallel field data: • <Original scripts> • 江戶 / 田中優子編. • Contributors:田中優子, 1952- • Format: Book • Language: Japanese • Published: 東京 : 作品社, 1998. • Series: 日本の名随筆. 03 别卷 ; 94 • <Paired w/OS first> • 江戶 / 田中優子編. • Edo / Tanaka Yūko hen. • Contributors: 田中優子, 1952- • Tanaka, Yūko, 1952- • Format: Book • Language: Japanese • Published: 東京 : 作品社, 1998. • Tōkyō : Sakuhinsha, 1998. • Series: 日本の名随筆. 03 别卷 ; 94 • Nihon no meizuihitsu. 03 Bekkan ; 94 We would like to choose our preference of display script here. For example, <Original scripts> 江戸 By: 野村兼太郎, 1896-1960. Published: 1942 Format: Book, Electronic Resource 江戶 の 翻訳家たち By: 杉本 つとむ, 1927- Published: 1995 Format: Book, Electronic Resource Jan 2010
Language/Script of Interface _____________________________________________________ OCLC’s brief record display Interface easily flipped to one of several languages Jan 2010
Language/Script of Interface_____________________________________________________ OCLC’s detailed record display with Japanese language interface Jan 2010
Language/Script of Interface OCLC WorldCat.org does localization of labels and instructions as well as localization of mapped facet values. Examples here in Chinese.
Language/Script of Interface_____________________________________________________ Jan 2010
Language/Script of Interface & Text Directionality_____________________________________________________ Jan 2010
Sorting of Results_____________________________________________________ 江戸文学俗信辞典 Edo bungaku zokushin jiten 江戸文学地名辞典 Edo bungaku chimei jiten 江戸文学辞典 Edo bungaku jiten 江戸文様辞典 Edo mon’yo jiten Jan 2010
Sorting of Results_____________________________________________________ Also note bi-directional text Jan 2010
Sorting within result sets: Options to Consider_____________________________________________________ For multiple languages sharing a script, e.g. Chinese ideographs, Arabic, Hebrew, or Latin, how would the users prefer to see the result sets sorted? We consider here the Chinese & Arabic cases… Jan 2010
Sorting within Result Sets: Options to Consider_____________________________________________________ Sorting of results returned in Chinese script— Three sort strategies: (a) sort by Romanized equivalents; (b) sort by pronunciation; or (c) sort by radical-stroke? Jan 2010
Sorting within Results Sets:Arabic script_____________________________________________________ How to handle additional Arabic-script characters in use for languages such as Persian, Kurdish, and/or Urdu? ڤ (vah, derived from ﻑ, fah) پ(pah) ﭺ (chah, derived from ج , ǧim) گ (gaf) ژ (zāī, derived from ز, zayin) Jan 2010
Discussion User Needs and Expectations Jan 2010