Identification Work Package 2 of Telematics for Libraries project BIBLINK (LB 4034)

Title page Table of Contents Previous - Next

6. Evaluation of the Identification Schemes

6.1 Evaluation introduction

In this section a brief overview and comparison of the different schemes described in section 5 is given. Each of the schemes is then compared and evaluated separately according to the requirements. A conclusion is drawn for each identification scheme.

6.2 Brief overview of how the identification schemes meets the requirements

The comparison of the different identification schemes can be found in table 2. Each identification scheme is compared to the requirements set up in section 4.3.

For explanation of the columns and a complete listing of the requirements, see section 4.3:

Id-scheme	1. Coverage	2. Authorised	3. Standard	4. Uniqueness	5. Persistence	6. Extensibility	7. Human readable	8. Transportable	9. Validation
ISSN	Only serial items, medium independent	Under control of a central agency.	Yes. ISO 3297, ANSI/NISO Z39.9	Yes	Yes	No	Yes	Yes		Yes
ISBN	All monographs, medium independent.	Under control of a central agency.	Yes. ISO 2108	Yes in principle. Depending on the publishers.	Yes.	No.	Yes	Yes	Yes
SICI	Serial issues and articles, medium independent.	Under control of a central agency.	Yes. ANSI/NISO Z39.56	Yes in principle.	Yes.	Yes.	Yes	Yes	Yes.
PII	Items within serial or monographic titles, medium independent.	No responsible authority for assignment. Assigned by the publishers.	No. The ISSN or ISBN included are both ISO standards.	No.	Yes.	Yes.	Yes.	Yes	Yes
DOI	On-line and off-line(?) documents.	Under control of a central agency.	No. Under development.	Yes	Yes	Yes	Yes	Yes	Depending on external resolution.
URN	All Internet documents.	Under control of a central agency.	No. Under development.	Yes.	Yes.	Yes.	Yes	Yes	Depending on external resolution.
PURL	All World Wide Web documents.	No	No.	Yes.	Uncertain. Depending on the resolver service.	Yes.	Yes.	Yes	Depending on external resolution.

Table 2: Comparison of identification schemes.

6.3 Evaluation of the identification schemes

Each of the identification schemes is evaluated individually on the basis of the investigations made in section 5 and a conclusion is drawn on whether or not the identification scheme meets the requirements set up for a BIBLINK identifier.

6.3.1 Evaluation of ISSN (International Standard Serial Number)

The ISSN can only be used for serials, including electronic serials, both off-line and on-line. There might be restrictions regarding "the size" of the serial, or rather the importance of the publisher. World-wide, ISSN is not assigned to small pamphlets or leaflets published by, for instance small special interest groups or to electronic serials that are regarded as an advertisement (e.g. that only consist of abstracts or a table of contents) for the serial on paper. This is not a major problem within the BIBLINK scope, since most of these documents would be excluded from the national bibliography anyway.

The ISSN is an ISO standard (ISO 3297) and under control of both an international and by national or regional agencies.

An ISSN will be both globally unique and persistent. The same content issued on different media (e.g. paper, CD-ROM and World Wide Web) will, according to the guidelines, be assigned different ISSNs. The identification scheme may not be extended but recently a new medium code has been implemented in the ISSN Register and a new linking field has been added in the ISSN format to link the different medium editions.

The ISSN may easily be transcribed for citation purposes and transported by Internet protocols. ISSNs could easily be validated locally.

The ISSN is widely used on all paper documents and is, to an increasing extent, also used on electronic documents.

6.3.1.1 ISSN conclusion

The ISSN meets all the requirements set up for a BIBLINK identifier except that it does not cover all documents in the BIBLINK scope. Section 4.1 Requirement introduction stated that it will not be possible to recommend only one identification scheme for BIBLINK. Based on this premise, it is not a significant problem that the ISSN does not cover all documents in the scope. The ISSN could be used for serials and other identifiers for non-serial documents.

Although the ISSN can be assigned to all electronic serials, the practice has varied widely in the past according to the policies of (the institutions hosting) ISSN centres in different countries. Some ISSN agencies will assign ISSNs to electronic serials only at the request of a publisher, while other agencies will go looking for new serials. Journals published, for instance, on paper, CD-ROM and on the Internet may be given three different ISSNs by some agencies, while other agencies will assign an ISSN only to the paper issue.

This might be a question of resources, not necessarily of principle. Basically the ISSN network has adopted a policy of assigning ISSNs also to all electronic serial publications. However several ISSN centres have less technical knowledge or resources or are part of institutions which do not yet have definite policies concerning this issue. Some policies can, for example, be closely related to the national depository policies or laws.

The ISSN is widely used and is a very well known identifier to all traditional publishers. The same may not be the case for «new publishers» of electronic publications. Many of the «new publishers» are not true publishers but organisations who only publish, for instance, a news bulletin. They are not necessarily acquainted with the ISSN system.

It might be argued that it is irrelevant to assign ISSNs to on-line serials as an ISSN is primarily seen as a control number for ordering system or connected to other advantages for physical media, for instance benefiting from postal reductions. This will change when trading of on-line publications via the Internet becomes more widespread. On the other hand some publishers, for example in the academic sector, may want to use ISSN for prestige reasons - recognised paper journals have ISSNs, so should electronic journals.

In the new guidelines from the International ISSN/ISBN agency a more general need for an unique identifier is emphasised, e.g. securing bibliographic control and using identifiers as a control number in databases.

The new guidelines from the international ISBN/ISSN agencies, which include all electronic documents (both on-line and off-line), might encourage further use. With one common practice for all electronic documents, for both ISSN and ISBN, it might be easier for the national agencies to establish and enforce a national policy and practice. The spread of the ISSN for electronic publications will depend on the extent to which national practice can promote the use of the ISSN on new electronic serials, especially when it comes to spreading the use among «new publishers».

The ISSN is recommended as an identifier for BIBLINK.

6.3.2 Evaluation of ISBN (International Standard Book Number)

The ISBN was established to cover all printed monographs, but is now also used on «monographs» issued on microforms, audio- and videocassettes and on CD-ROMs and floppy disks. According to the new guidelines from the international ISBN agency on-line documents will also be included.

The ISBN is a recognised standard and the assignment of group and publisher identifiers is under the control of international and national agencies, while the assignment of ISBNs to individual documents is done by the publishers themselves. Assignment is under control of international and national guidelines.

In principle an ISBN is unique. The same content issued in different versions, e.g. on paper and on CD-ROM will, according to the guidelines, be assigned different ISBNs. The ISBN is also, in principle, persistent but ultimately it is the individual publisher who is responsible for ensuring that the same ISBN is not assigned to several documents and for not reusing the numbers. ISBNs could be misused by publishers and the only sanction against this breach of the rules is for the ISBN agency to refuse to issue any more ISBNs to the publisher or to refuse to take any books with reused numbers into any databases the agency may be responsible for (e.g. Whitakers Books in Print).

The identification scheme may not be extended, but may easily be transcribed for citation purposes and be transported by Internet protocols. ISBNs could easily be validated locally.

The ISBN is widely used on all paper documents and has also, to some extent, been assigned to off-line publications but, to date, not to on-line documents (though there are some exceptions).

The ISBN meets most of the requirements for a BIBLINK identifier. There might be a problem with the reuse of numbers, but this is not regarded as a major problem. The extensibility could become a problem if the large number of electronic documents on the Internet were to be allocated ISBNs.

To date, the ISBN has not been assigned to electronic documents to a significant degree. ISBNs are allocated to off-line multimedia and composite products that are sold in the book trade (e.g. CD-ROMs with manual). When traditional publishers issue multimedia they tend to give them ISBNs, the same is not the case when the publisher is a «new publisher» or when the document is published on the Internet.

The purpose of the ISBN has been to facilitate transactions, e.g. ordering, and this has, up until now, not been appropriate for on-line products. Online monographic publications on the Internet are still mainly public domain publications and are not taking part in the traditional trading and distribution chains. One problem is that, up until now, Internet technology has not offered the possibility of secured online transactions. Some of the ISBN agencies are strategically situated in the centre of the trading and distribution chain, and this is why Internet publications might not even come to the attention of ISBN agencies. The ISBN agencies will probably become more active on the Internet when the trading of online publications via the Internet becomes more widespread.

In the new guidelines from the International ISSN/ISBN agency a more general need for a unique identifier is emphasised, e.g. securing bibliographic control and using identifiers as a control number in databases.

The new guidelines from the international ISBN/ISSN agencies, which include all electronic documents (both on-line and off-line), might encourage their use further. With a single common practice for all electronic documents, for both the ISSN and the ISBN, it might be easier for the national agencies to establish and enforce a national policy and practice. The spread of the ISBN on electronic publications will depend on the extent to which national practice can promote the use of ISBN on new electronic documents, especially when it comes to spreading the use among «new publishers».

In some cases, on-line documents are likely to change on a fairly regular basis and consequently ISSN may be more appropriate.

The ISBN is recommended as an identifier for BIBLINK.

6.3.3 Evaluation of SICI (Serial Item and Contribution Identifier)

The use of the SICI is limited to serials. A SICI can be used to identify both a complete document and items in a document, e.g. articles. The SICI may also be used for electronic documents, as long as they contain a location number or an enumeration. This will exclude a large proportion of the documents in the BIBLINK scope - all non-serial documents. The recent completion of the BICI will obviously solve this problem.

The SICI is an recognised international standard. The assignment of a SICI to a document is done by the individual publishers or others in need of a SICI identifier for a document.

The SICI will be both unique and persistent. There is a small theoretical possibility that two contributions can have identical values, but tests indicate that the duplicate values occur only once per million contributions. Even so the SICI, as a whole, will be unique. There is also a possibility that a contribution might be defined by more than one SICI. A SICI may be derived from different sources and depending on the information (more or less complete) available when the SICI is constructed, the same item might be given different contribution identifiers. This is not regarded as a problem and is not very likely to occur.

The latest version of the scheme is extended to include contributions other than articles, e.g. tables of contents. In principle the SICI could be extended similarly for other reasons, though this is not a proposition at the moment.

The SICI may easily be transcribed by humans for citation purposes and be transported by Internet protocols. SICIs can easily be validated locally however, local comparison may be more difficult.

The SICI is widely used, not only by publishers but also by other members of the bibliographic community. SICIs are primary an aid to finding existing published articles or issues.

The SICI meets most of the requirements set up for a BIBLINK identifier. SICI does not cover all documents in the BIBLINK scope.

Section 4.1 Requirement introduction stated that it will not be possible to recommend only one identification scheme for BIBLINK. Based on this premise, it is not a problem that the SICI does not cover all documents in the scope. The SICI could be used on all serial documents and other identifiers on non-serial documents, for instance the BICI.

The SICI includes an ISSN and can consequently only be assigned to documents which are already assigned (or can be assigned) an ISSN. The use of SICIs on electronic documents will therefore depend on the policy for assigning ISSN to documents (See 6.3.1 ISSN). This is not regarded as a problem in the BIBLINK context, since all documents in the scope could be assigned an ISSN.

The SICI is recommended as an identifier for BIBLINK.

6.3.4 Evaluation of PII (Publisher Item Identifier)

The PII may be used on all electronic documents, both serial and non-serial. The identifier is designed to be used at the item level, i.e. articles in a serial or chapters in a book and not at the medium level, i.e. on complete documents. The PII is therefore only appropriate for identifying documents in the BIBLINK context when single articles or chapters are presented to the national library for bibliographic recording and/or deposit.

The PII is not under control of any responsible authority and is not a recognised international standard. The identifier is assigned to the documents by the publishers themselves. This implies that the uniqueness and the persistence of the identifier depends on the individual publishers.

Even so, the PII is «guaranteed» to be both globally unique and persistent. However, it is not unique in the way described in section 4.3: «A document published in several versions (e.g. on CD-ROM and World Wide Web) should have separate unique identifiers». An article published on, for example, paper and the Internet will carry the same PII (carrying the ISSN of the original document). An important point about the PII is that it identifies articles independently from their packaging unit. Consequently a PII may have one ISSN prefix and appear in a publication with a different ISSN.

The identifier may, in principle, be extended to identify components (tables, graphics) and manifestations (e.g. SGML) of an item. The identifier may easily be transcribed by humans for citation purposes and be transported by the common Internet protocols. The PII can be validated locally.

The PII is a «new» identifier, adopted for all articles published by the publishers involved in the PII initiative from 1996 onwards, by Springer and some other primary publishers as well as by their secondary databases (e.g. Chemical Abstracts). The use of PII is encouraged by the initiators and is expected to spread. The PII is primarily aimed at documents of interest to scientific publishers.

The PII fails to meet the most important requirements set up for a BIBLINK identifier. The fact that the same identifier will be assigned to items (e.g. articles) issued on different media is a major objection to the identification scheme. An important demand of a BIBLINK identifier is that it will identify different versions of a publication separately.

Another objection is that it is assigned at the item level and not at the medium level. The PII could therefore not be used to identify complete documents, which will be the majority of the documents dealt with in the BIBLINK context. In the main it will be complete books or serials that will be presented to the national library for bibliographic recording and/or deposit. Within a serial contents database an identification scheme like PII could be interesting, but this interest is outside the BIBLINK scope. If and when publishers start to publish articles on demand or to provide a database with full-text articles, an identifier at the item level might be useful.

The PII is not recommended as an identifier for BIBLINK.

6.3.5 Evaluation of DOI (Digital Object Identifier)

The DOI can be used on all types of electronic documents. The DOI is designed for on-line documents and there are, at the time of writing, some concerns about its use for off-line products. DOIs can also be used for new products beyond traditional print and online equivalents to print, for example software plug-ins which create chemical models that can be analysed and rotated.

The DOI may be used to identify any electronic publication at any level of granularity, both at the item level and at the medium level. A journal can have a DOI, each of the articles it contains may also have a DOI and each of the pictures in those articles may have a DOI.

The DOI is not as yet a recognised standard and is still under development. According to the plans, the identifier will be under control of a central agency, which will provide identifiers to publishers and quality control for the entire DOI system. The DOI is a «new» initiative, the project only started in 1996, but the underlying Handle System is already available and it seems likely that the DOI system will be widely adopted by the publishing industry. A demonstrator system is now available and it is planned that a working prototype will be unveiled at the Frankfurt Book Fair in October 1997.

The DOI is guaranteed to be globally unique, persistent and extensible. The identifier may easily be transcribed for humans for citation purposes and be transported by the common Internet protocols.

The DOI will depend on an external resolution service for validation. At present most library systems are programmed to validate and compare identifiers like ISBN and ISSN. If a DOI is used the library system could not perform these controls locally but would have to make contact with an external resolution service for comparison and validation. This would require adaptation of the existing library systems and well functioning and stable resolution services.

Except for not being a standard, the DOI meets most of the requirements set up for a BIBLINK identifier and also, importantly, the DOI is specifically developed to identify electronic documents. In addition, the DOI will act as a resolution service for electronic documents, in a similar way to the PURL system and what is planned in URN systems. Therefore DOIs may also solve the «moving URLs» problem. Presumably the DOI system may eventually become part of a wider URN system.

The DOI is recommended as an identifier for BIBLINK. An important question is how widespread the use of the identifier will be, particularly in Europe. Since the DOI will be a very useful identifier for BIBLINK, the national libraries should encourage the development of the identifier.

6.3.6 Evaluation of URN (Uniform Resource Name)

The URN is being developed to identify any electronic resources on the Internet. Off-line documents, e.g. CD-ROMs or floppy disks could not be assigned a URN.

The URN will be under the control of an authorised partner, a «Namespace identifier» who will be responsible for a correct assignment of URNs.

According to the requirements, the URN will be both globally unique, persistent and any scheme defined as a «Namespace» must be extensible. A URN can be transported by Internet protocols and easily be transcribed by humans for citation purposes.

The URN will depend on an external resolution service for validation. Currently most library systems are programmed to validate and compare identifiers like ISBN and ISSN as they are entered into the system. If a URN is used, the library system could not perform these functions locally, but would have to make contact with an external resolution service for comparison and validation. This would require adaptation of the existing library systems and well functioning and stable resolution services.

The URN is not a standard as yet and is still under development.

The development of the URN is closely connected to the use of URLs. A URL (Uniform Resource Locator) is the most common way of providing access to an Internet resource and is used in the hypertext links in World Wide Web documents. URLs are also recorded in bibliographic catalogues, sometimes as a hypertext links to the document itself. A URL is not a unique and persistent identification for an electronic document . The URL is both the name and the address (location) of the document. If the document is moved, the URL will change and the document may be difficult to locate. Duplicate copies of the same content will have different URLs. This is a general problem for all users of Internet resources.

There is reason to believe that the URN will be a part of a resolution service, which will link persistent and unique URNs to one or more current URLs. A URC (Uniform Resource Characteristics) is one way of developing a resolution service. A URC could contain the URN and at least one URL.In addition, a URC might contain bibliographic and other information about the resource, so called metadata. This would solve the «moving URLs» problem and makes it very likely that URNs will be widely adopted on the Internet

The fact that other identification schemes, for instance ISSN or SICI, may be recognised as a «Namespace» could make the URN easy to adopt for publishers. They can continue to use an identification scheme, which is useful for other reasons than location, e.g. ordering or management rights and at the same time have a URN assigned to their documents.

The URN meets most of the requirements for a BIBLINK identifier, except that it can not be used to identify off-line documents. Section 4.1 Requirement introduction stated that it will not be possible to recommend only one identification scheme for BIBLINK. Based on this premise, it is not a significant problem that the URN does not cover all documents in the scope. The URN could be used for all on-line documents and other identifiers could be used for off-line documents.

There is the concern that the identifier does not «exist» as yet, but when (if) it does the identifier will be very suitable for identifying all on-line documents in the BIBLINK scope. The success of the URNs may be dependent on a well organised resolution service.

The URN is recommended as an identifier for BIBLINK.

6.3.7 Evaluation of PURL (Persistent Uniform Resource Locator)

The PURL may be used to identify any World Wide Web document, but can not be assigned to off-line documents. A PURL can be assigned at both the document level and at the item level, i.e. to individual parts of the document.

The PURL is not a standard and will not become one. The PURL service is set up as a short term solution and will be replaced by URNs when this identification scheme is available.

There is no international or national authority responsible for assigning PURLs. The resolver service (OCLC or other) will be responsible for the linkage between PURLs and URLs, but publishers may allocate PURLs for their own documents.

PURLs are unique and easily extensible. Their persistence is dependent on the lifetime of the resolver service. However it is expected that organisations that are willing to set up a PURL resolution service show commitment on a long term basis. PURLs can be easily transcribed by humans for citation purposes and transported by the common Internet protocols

PURLs depend on an external resolution service for validation. Currently most library systems are programmed to validate and compare identifiers like ISBN and ISSN locally, as they are entered into the system. If a PURL is used, the library system could not perform these functions locally, but would have to make contact with an external resolution service for comparison and validation. This would require adaptation of existing library systems and well functioning and stable resolution services.

Important requirements for a BIBLINK identifier are not met by the PURL. The PURL initiative is a short term solution and is not be recommended as an identifier for BIBLINK.

A PURL may still be of interest to the national libraries as a temporary solution for keeping track of Internet resources until the URN is available. If documents recorded in the national bibliography are assigned a PURL, many of the problems concerning electronic document and catalogue maintenance will be solved.

Next	Table of Contents