NISO RP-2005-2003 Metasearch Initiative - Search and Retrieval Citation Level Data Elements《搜索和检索引用级别数据元》.pdf

《NISO RP-2005-2003 Metasearch Initiative - Search and Retrieval Citation Level Data Elements《搜索和检索引用级别数据元》.pdf》由会员分享，可在线阅读，更多相关《NISO RP-2005-2003 Metasearch Initiative - Search and Retrieval Citation Level Data Elements《搜索和检索引用级别数据元》.pdf（6页珍藏版）》请在麦多课文档分享上搜索。

1、NISO RP-2005-03 NISO Metasearch Initiative Search and Retrieval Citation Level Data Elements A Recommended Practice of the National Information Standards Organization Standards Committee BC / Task Group 3 Version 1.0 September 13, 2005 Published by the National Information Standards Organization Bet

2、hesda, MD 2005 NISO 1 Summary The NISO Metasearch Initiative, Task Group 3/SubGroup 3, on Required Citation Metadata has discussed the issues around citation metadata and its relation to metasearch. Citation references have been devised in a paper world, assuming page numbers and enveloping journals

3、 and publishers. But searchers will use metasearch engines to search, find, and retrieve individual articles. There are a number of extant issues that must be addressed to allow smooth and seamless metasearching across multiple resources. The Google Scholar approach is to access the full-text conten

4、t of all available journals and provide a heterogeneous data store. Unfortunately, for researchers, they need fine tuning of their search experience with relevant metadata so as not to be swamped by irrelevant references. Our proposed approach is simply to have a consistency in the format and conten

5、t of citation metadata. Issues Inconsistent Citation Styles The reference styles for citations tend to differ according to discipline. There are tens, if not hundreds of styles. As an example, one vendor has seventeen citation formats across twelve databases. The ISO and NISO standards are not in th

6、emselves a sufficient guide to all the variations. From the Dublin Core Metadata Initiative Citation Working Group, we get the following list of variations: The order of elements (especially elements such as initials) The mandatoriness of elements (e.g. many chemistry styles leave out the article ti

7、tle, but biology and medicine wouldnt) The punctuation between the elements Capitalization. E.g. of titles - some styles use “title case“ (i.e. initial capitals for all main words) and some use “sentence case“ (i.e. initial capitals for first word and proper nouns and adjectives only) Acceptable abb

8、reviations (especially regarding journal title abbreviations, but also element indicators such as “chapter/chap/ch“, “editor(s)/edited by/ed(s)“, “edition/edn/ed“ Character formatting (i.e. what goes in italic, bold, etc.) Refer to http:/epub.mimas.ac.uk/DC/citstyles.html for more discussion and a l

9、ist of citation styles. One of the reasons behind this plethora of styles is that data vendors purchase data from different publishers, each using potentially different styles. Complex Technology Required Due to the wide and varying citation formats returned by various vendors, metasearch engines mu

10、st choose how to parse each citation. With “random” fields, even the parsed results are unreliable and inconsistent, oftentimes producing bad OpenURLs which can make it difficult for users to get to the full-text or article that was originally published. Vendor Branding Vendors and publishers desire

11、 to maintain their branding and identity in results sent to users, even after being massages by a metasearch engine. Either a vendor produces a proprietary OpenURL that will only point back to their own sources, or a vendor or publishers reference is lost from the metadata. The vendor wants more exp

12、osure, renewed subscriptions, and possibly pay-per-view of full-text. Mapping of Metadata One issue that causes confusion and difficulty in de-duping records is the process in which multiple metadata items get placed into databases. A typical scenario goes as follows: the primary publisher creates a

13、 human readable citation field; the human readable citation field is dumped into a single database field; and the record in the database is sold to an aggregator. Since many different formats may 2005 NISO 2 be managed by one citation aggregator, it is difficult to tell which format they used for ea

14、ch citation. When the record is searched, it may be displayed as created by the publisher and not the authors. Requirements The requirements to enable effective and seamless metasearch across multiple databases and resource types are surprisingly simple. There are basically two audiences to the resu

15、lts of a metasearch: a metasearch engine, and the end-user. The combined minimum requirements end up being as follows: Minimum metadata to allow a metasearch engine to compare results from multiple resources: Unambiguous metadata Enough to be able to Sort/Merge/Dedupe (OpenURL) Display (Brief/Full)

16、minimum for the user Produce OpenURL/Link Ranking: Need searched fields: Subject/Description/Abstract To create a “Brief” Record, you need, at a minimum: Genre what “type“ of item is it? Creator who created the original article? Title how is this article referred to? ID what ID(s), such as PII, SICI

17、, DOI, etc., is this article known by? Context what enveloping publication or proceeding, etc., is this article found in? To create a “Full Display“ Record, and to enable ranking and full-text analysis of the metadata, you need: Subject for cataloged subject headings Description some text describing

18、 what this item Proposed Solution A detailed table describing the minimum data elements needed for citation metadata follows this summary; an XML version of the table is available on the NISO Metasearch Initiative website (http:/www.niso.org/committees/MS_initiative.html). This set is taken extensiv

19、ely from Dublin Core 0.1, qualified for citations from the citation working group, however, it adds the descriptive components needed for “Full Display“ and text analysis done by metasearch engines. A quick overview follows. As expected, it closely matches the Requirements listed above. “genre“ elem

20、ent that describes WHAT kind of object we have an “authors“ field, as in OpenURL “titles“ field that has Journal Title and Article Title “dates“ field that has the date of publication, and other chronological information if present “context“ field that gives volume, issue, pages, etc. “citationID“ f

21、or ISBN, ISSN, SICI, etc. “publisher“ field, if available “fulltextURI“ to point to the full-text, if available 2005 NISO 3 For full display information, add the following. (If the information is requested by a metasearch server that is doing independent ranking of results, then this information is

22、highly recommended to aid in the ranking of results.) “description“ as in Dublin Core, for description or abstract “subject“, as in Dublin Core, for subject headings “vendorData“ to include, in “free form“ with a schema pointer, whatever else they want to add (This allows vendors to preserve brandin

23、g.) Links For comparison and related links, here are other, similar standards, and a few discussions of interest: Dublin Core Metadata Initiative Citation Working Group http:/dublincore.org/groups/citation/ Guidelines for Encoding Bibliographic Citation Information in Dublin Core Metadata http:/www.

24、dublincore.org/documents/dc-citation-guidelines/ IMS Resource List Interoperability (RLI) Information Model, e-Learning metadata http:/www.imsglobal.org/rli/rliv1p0/imsrli_infov1p0.html Digital Objects Requirements: Metadata, California Digital Library http:/www.cdlib.org:8081/inside/diglib/guidelin

25、es/mdreqs.html#mdguidelines XBib - Bibliographies and Citations for XML http:/ RIS Format Specifications (EndNote) http:/ Identifier Encoding Schemes http:/epub.mimas.ac.uk/DC/citids.html ANSI/NISO Z39.56-1996 (R2002), Serial Item and Contribution Identifier (SICI) http:/www.niso.org/standards/stand

26、ard_detail.cfm?std_id=530 Marc Proposal No:2003-03, Definition of Data Elements for Article Level Description, Library of Congress http:/www.loc.gov/marc/marbi/2003/2003-03.html Cameron, Robert D., Towards Universal Serial Item Names, School of Computer Science, Simon Fraser University, CMPT TR 97-1

27、6 http:/www.cs.sfu.ca/pub/cs/TR/1997/CMPT97-16.html Green, B., and Bide, M., Unique Identifiers: a brief introduction http:/www.bic.org.uk/uniquid.html Registry for the OpenURL Framework - ANSI/NISO Z39.88-2004 http:/www.openurl.info/registry/ Metasearch Search and RetrievalCitation Level Data Eleme

28、ntsv 1.0Descriptioncitation The required root element “citation“ contains child elements that are used to express properties of serial publicationsgenre Genre of the document. Legitimate values for the “genre“ element are: (1) “journal“: for a serial publication issued in successive parts (2) “issue

29、“: for one instance of the serial publication (3) “article“: for a document published in a journal. (4) “conference“: for a record of a conference that includes one or more conference papers and that is published as an issue of a journal or serial publication (5) “proceeding“: for a single conferenc

30、e presentation published in a journal or serial publication (6) “preprint“: for an individual paper or report published in paper or electronically prior to its publication in a journal or serial (7) “book“ for monographs (8) “bookitem“ for parts of a book, such as a chapter (9) “unknown“: use when t

31、he genre is unknown.creator The “creator“ element contains child elements that are used to express authorship of an individual item in a publication. The “creator“ element is not repeatable, it contains all authors, and allows for the indication of the position of the author in the publications list

32、 of authorsattr:rank An integer indicating the position of the author in the publications list of authors , e.g. “1“ for first author, “2“ for second author, etc.author The person primarily responsible for creating the intellectual content of the resourceaulast The authors family name. This may be m

33、ore than one word. In many citations, the authors family name is recorded first and is followed by a comma, i.e. Smith, Fred James is recorded as “aulast=smith“aufirst The authors given name or names or initials. This data element may contain multiple words and punctuation, i.e. “Fred F“, “Fred Jame

34、s“auinit The authors first and middle initials.auinit1 The authors first initial.auinitm The authors middle initial.ausuffix The authors name suffix. Qualifiers on an authors name such as “Jr.“, “III“ are entered here. i.e. Smith, Fred Jr. is recorded as “ausuffix=jr“au The authors full name, i.e. “

35、Smith, Fred M“, “Harry S. Truman“aucorp Organization or corporation that is the author or creator of the book, i.e. “Mellon Foundation“title The “titles“ element contains child elements that are used to express the fully qualified title of an individual article in a serial publication. The “titles“

36、element is not repeatable, it contains the journal or abbreviated journal title, article titleArticle titleEither the Journal Title “jtitle“ or abbreviated journal title, “stitle“ must be supplied. However, the data vendor may supply both if they are available.jtitle Journal title. Use the most comp

37、lete title available, e.g. “journal of the american medical association“. Abbreviated titles, when known, are provided in the “stitle“ element.stitle Abbreviated or short journal title. This is used for journal title abbreviations, e.g. “J Am Med Assndate The “date“ element contains child elements t

38、hat are used to express the fully qualified date, time, or season of when an individual article in a serial publication was published. The “date“ element is not repeatableisoDate Date of publication in ISO 8601 form YYYY, YYYY-MM or YYYY-MM-DDchron Indications of chronology in a non ISO8601 form (li

39、ke “Spring“ or “1st quarter“) should be carried in this element; the element content is not normalized. Where numeric ISO8601 dates are also available, they should be provided in the “date“ element. As such, a recorded date of publication of “Spring, 1992“ becomes “date=1992“ and “chron=spring“. Chr

40、onology information can also be provided in the “ssn“ and “quarter“ elementsElementatitle(c) 2005 NISO4Metasearch Search and RetrievalCitation Level Data Elementsv 1.0DescriptionElementseason Season (chronology). Legitimate values are “spring“, “summer“, “fall“, “winter“quarter Quarter (chronology).

41、 Legitimate values are “1“, “2“, “3“, “4“context The “context“ element contains child elements that are used to express the fully qualified location of an individual article within a serial publication. This is the volume, issue, page number, etc. Since different publications may or may not have any

42、 one of these child elements, they are not specified here other than requiring a context element for citation level reference.volume Volume designation. Volume is usually expressed as a number but could be roman numerals or non-numeric, e.g. “124“, or “VI“.4“part Part can be a special subdivision of

43、 a volume or it can be the highest level division of the journal. Parts are often designated with letters or names, e.g. “B“, “Supplement“issue This is the designation of the published issue of a journal, corresponding to the actual physical piece in most cases. While usually numeric, it could be no

44、nnumeric. Note that some publications use chronology in the place of enumeration, i.e. Spring, 1998.spage Start, or first page number of a start/end (spage-epage) pair. Note that pages are not always numeric.epage Second (ending) page number of a start/end (spage-epage) pairpages Start and end pages

45、 in the form “startpage-endpage“. This field can also be used for an unstructured pagination statement when data relating to pagination cannot be interpreted as a start-end pair, i.e. “A7, C4-9“, “1-3, 6“artnum Article number assigned by the publisher. Article numbers are often generated for publica

46、tions that do not have usable pagination, in particular electronic journal articles, e.g. “unifi000000090“. If article numbers are identifiers that follow a URI Scheme such as “info:doi/“ the information should be provided in the Identifier Descriptor of the ContextObject, not in this “artnum“ eleme

47、nt. Likewise, if articles are identified by means of a registered URI Scheme such as the http scheme, the information should be provided in the Identifier Descriptor of the ContextObjectThe “id“ element contains child elements that are used to describe the standard ID assigned to the journal, book,

48、serial, etc. It may be the ISBN, ISSN, EISSN, CODEN, or SICI.Enumeration values:ISSN: International Standard Serial Number (ISSN). ISSN numbers may contain a hyphen, e.g. “1041-5653“EISSN: ISSN for electronic version of the journal. Although there is no distinction by format in the assignment of ISS

49、Ns, some bibliographic services now carry both the ISSN for the paper version and a separate ISSN for the electronic version. This data element is included here to allow expression of both types of ISSN numbersISBN: International Standard Book Number (ISBN). The ISBN is usually presented as 9 digits plus a final check digit (which may be “X“), e.g. “057117678X“. ISBN numbers may contain hyphens, e.g. “1-878067-73-7“SICI: