Format Conversion Feasibility Work Package 4 of Telematics for Libraries project BIBLINK (LB 4034)

Title page Table of Contents Previous - Next

6.4 Mapping SSSH to UNIMARC

Many publishers' formats are based upon implementations of SGML [8]. SGML (ISO 8879: 1986) is a language for expressing context free document description [9]. It is realised by means of an application program called a Document Type Definition or DTD. DTDs exist for many different classes of document and DTDs for books, articles and serials are published in ISO 12083: 1994 [10]. Because publishers use SGML applications it will be useful to look at mapping the information found in one of these formats to one of the MARC records used by national bibliographic agencies: in this report UNIMARC.

6.4.1 Simplified SGML for Serial Headers (SSSH)

Simplified SGML for Serial Headers (SSSH) is a SGML application for periodical headers. It was developed by Pira International for Book Industry Communication (BIC). SSSH is based on an established SGML Document Type Definition (DTD); the European Workgroup for SGML (EWS) Modular Application for Journals (MAJOUR) header which is used by a number of major European periodical publishers including Elsevier Science, Kluwer and Springer. SSSH is designed to encompass the needs of both the simpler MAJOUR application and the more complex Organization for Article Standards in Science (OASIS) data set [11].

Typically, SSSH would be used to mark-up a periodical header containing information on an article or periodical issue. To this end, the SSSH DTD defines data elements for two main subgroups: the issue and the article. This focus proves problematic for the creation of any relevant MARC record which would primarily be interested in the bibliographic history of a periodical. This initial mismatch between two relatively complex formats leads to significant problems in mapping between the two different formats. These problems remain when even a minimal CIP type record is required.

6.4.2 Mapping to UNIMARC

Two tables have been produced for a SSSH mapping to UNIMARC, one for issue-level information, the other for article-level information.

Issue level mapping.

Table II. Mapping tables for issue-level SSSH - UNIMARC:

6.4.3 Comments on SSSH issue level information mapping to UNIMARC

SSSH Issue information includes details like the periodical title and publisher, the most relevant information for a minimal level description of a periodical in UNIMARC. For example, the following fragment of an SSSH record would describe one issue of the weekly (printed) British Medical Journal:

Example SSSH issue-level record:

Using the above SSSH to UNIMARC mapping the following UNIMARC issue level record could be created:

Or if treated as a series:

In addition to not containing some mandatory UNIMARC fields (e.g. 100), the resulting UNIMARC record will also omit the following important bibliographic information:

• Regularity of issue
• First date of publication
• Publication history
• Physical details

The detail in the SSSH application is actually at the article level rather than the journal level. This means that SSSH would be a poor source of information even for a minimal level CIP type bibliographic record for a journal. On the other hand, SSSH information might possibly be in a good position to create an adequate article-level UNIMARC record.

6.4.4 Article level mapping

Table III: Mapping tables for article-level SSSH - UNIMARC

6.4.5 Example SSSH article-level record:

<header>
<issue>
<pinfo>
<pnm>American Mathematical Society
<loc>Providence, Rhode Island

<jinfo>
<jtl>Mathematics of Computation
<issn>0025‐5718
<pubinfo>
<vid>66
<iid>218
<cd>199704
<artcon>
<genhdr>
<artinfo>
<ppct 10>
<ppf>465
<ppl>476

<tig>
<atl>A posteriori error estimate for the mixed finite element method
<aug>
<au><fnms>Carsten<snm>Carstensen</au>
<aff>Mathematisches Seminar, Christian‐Albrechts‐Universität zu Kiel, Ludewig‐Meyn‐Str. 4, D‐24098 Kiel, Germany
<abs>
<p>A computable error bound for mixed finite element methods is established in the model case of the Poisson‐problem to control the error in the H(div,ω) xL2(ω)‐norm. The reliable and efficient a posteriori error estimate applies, e.g., to Raviart‐Thomas, Brezzi‐Douglas‐Marini, and Brezzi‐Douglas‐Fortin‐Marini elements
<kwdg>
<kwd>Mixed finite element methods, a posteriori error estimates, adaptive algorithm
</header>

It is a bigger challenge to create an UNIMARC record for this article-level header than for the issue-level headers discussed above. Using some ingenuity and the above mappings, the following record could be created:

200 1#$aA posteriori error estimate for the mixed finite element method$f Carsten Carstensen
210 ##$aProvidence, Rhode Island$cAmerican Mathematical Society$d1997
215 ##$a10 p.
225 1#$aMathematics of Computation$vvol. 66$vno. 218$x0025-5718
330 ##$aA computable error bound for mixed finite element methods is established in the model case of the Poisson-problem to control the error in the H(div,W) xL2(W)-norm. The reliable and efficient a posteriori error estimate applies, e.g., to Raviart-Thomas, Brezzi-Douglas-Marini, and Brezzi-Douglas-Fortin-Marini elements
610 0#$a Mixed finite element methods$aa posteriori error estimates$aadaptive algorithm
700 0#$aCarsten Carstensen

Again, this is not a valid UNIMARC record as it is missing the mandatory 001 and record label/100 fields. Fairly adequate descriptive UNIMARC records can be collected from SSSH but the records are still not valid without further intervention, probably by a human cataloguer.

6.4.6 Conclusions

An SGML DTD like SSSH allows the structured retrieval of quite a lot of detailed information. However, because SSSH exists to provide information at article level it lacks the information on frequency of issue and bibliographical history which would be useful for even a minimal UNIMARC record. Simple CIP-like records could be produced from SSSH which would contain just periodical title, subtitle, publisher, place of publication and ISSN, but as more than one SSSH record is likely to be extant for each periodical (one for each article produced) each ISSN should be checked to ensure that duplicate records are not being created. At the article level fairly full UNIMARC records could be produced, but as with the other mappings in this report some human intervention would be required to ensure that all the mandatory fields in UNIMARC are filled in correctly.

It seems that as the SSSH format is more suitable for serial description at the article level, rather than the issue level then BIBLINK will need to consider an alternative format for use at the issue level. Possibilities are Dublin Core or the format used in the ISSN database. This decision needs to take into consideration the findings of WP3 as regards the formats serial publishers use.

In themselves the various MARC formats are not well adapted for description of serials. BIBLINK must consider our approach to serial description in terms of the impact of adapting the existing MARC formats for local use, and how far we wish to formulate recommendations to the MARC standards making bodies.

References

[8.] Goldfarb, C.F. The SGML handbook. New York: Oxford University Press, 1991.

[9.] Wright, H. SGML Frees information. Byte, June 1992, 279.

[10.] ISO 12083: 1994(E). Information and documentation - electronic manuscript preparation and markup. Geneva: ISO, 1994.

[11.] Pira International. Simplified SGML for Serial Headers (SSSH). 1996.