Re: [CODE4LIB] Multiple ISBNs in COInS?
I agree; issn is not an identifier for an article. But in general, a resolver should be smart enough to know what serial is meant even if a variant issn is supplied. To prevent multiple searches, the resolver has to know how a title is referenced in the target source. This requires precalculation using a service or data file like xISBN that included ISSNs. However, it is important to keep in mind that sources such as the library catalog sometimes require multiple ISSNs to retrieve all holdings data unless this information is combined before it is loaded into the resolver knowledgebase. Between cataloging rules that influence how serials are issued (specifically, the practice known as successive entry cataloging which spreads individual titles across multiple records because of piddly variations in issues) and things that occur at the publishing end of things, many journals are known by multiple ISSNs. Practices like these are not user friendly -- even reference librarians don't seem to understand them -- so database providers typically combine all the issues so they can be considered part of one unit. Vendor provided data about such titles will likely include only one of these ISSNs (most likely, the most recent one, but that is not guaranteed). Unlike vendors, catalogers can be counted on to spread the holdings statements across multiple records and ISSNs if the cataloging rules so prescribe. This may sound like cataloging minutia, but this dynamic affects a number of very popular titles. Resolving only one ISSN could easily lead people to think an issue they need is not available when it is on hand. kyle -- -- Kyle Banerjee Digital Services Program Manager Orbis Cascade Alliance [EMAIL PROTECTED] / 541.359.9599
Re: [CODE4LIB] Multiple ISBNs in COInS?
It's an excellent point. The resolver's knowledgebase needs to know which issn a vendor has bundled content under, and ideally will be able to access that content no matter the issn/eissn in the OpenURL metadata. I'm thinking of a particular vendor that uses issn in the url syntax, but without a knowledge, it's hard to predict which issn is the one they use! On Feb 29, 2008, at 8:38 AM, Kyle Banerjee wrote: I agree; issn is not an identifier for an article. But in general, a resolver should be smart enough to know what serial is meant even if a variant issn is supplied. To prevent multiple searches, the resolver has to know how a title is referenced in the target source. This requires precalculation using a service or data file like xISBN that included ISSNs. However, it is important to keep in mind that sources such as the library catalog sometimes require multiple ISSNs to retrieve all holdings data unless this information is combined before it is loaded into the resolver knowledgebase. Between cataloging rules that influence how serials are issued (specifically, the practice known as successive entry cataloging which spreads individual titles across multiple records because of piddly variations in issues) and things that occur at the publishing end of things, many journals are known by multiple ISSNs. Practices like these are not user friendly -- even reference librarians don't seem to understand them -- so database providers typically combine all the issues so they can be considered part of one unit. Vendor provided data about such titles will likely include only one of these ISSNs (most likely, the most recent one, but that is not guaranteed). Unlike vendors, catalogers can be counted on to spread the holdings statements across multiple records and ISSNs if the cataloging rules so prescribe. This may sound like cataloging minutia, but this dynamic affects a number of very popular titles. Resolving only one ISSN could easily lead people to think an issue they need is not available when it is on hand. kyle -- -- Kyle Banerjee Digital Services Program Manager Orbis Cascade Alliance [EMAIL PROTECTED] / 541.359.9599
Re: [CODE4LIB] Multiple ISBNs in COInS?
It's an excellent point. The resolver's knowledgebase needs to know which issn a vendor has bundled content under, and ideally will be able to access that content no matter the issn/eissn in the OpenURL metadata. I'm thinking of a particular vendor that uses issn in the url syntax, but without a knowledge, it's hard to predict which issn is the one they use! Knowing which to put in the URL should be a knowledgebase issue since data indicating which ISSN to use as well as holdings should be available from the vendor. If the vendor doesn't provide this data, there's no choice other than to pass all searches to the DB, -- definitely a suboptimal solution. However, there's no rule that says a resolver can't learn these things on the fly (and if there were, this could be dismissed as a dumb rule that should be ignored). The first time someone connects to a title, multiple connections can be opened to determine preferred ISSNs as well as any holdings indicated on the target site. Once the good ISSN and holdings are remembered along with nonpreferred ISSNs which are still useful because they could be passed in a citation, it should only be necessary to open a single connection except in those cases where multiple connections are necessary. kyle -- -- Kyle Banerjee Digital Services Program Manager Orbis Cascade Alliance [EMAIL PROTECTED] / 541.359.9599
Re: [CODE4LIB] Multiple ISBNs in COInS?
I'm not sure, it seems to me that it could never be harmful, and could sometimes be helpful, for the CO generator to send all the information it has. Sure, a good link resolver _should_ be able to resolve any valid ISSN, it shouldn't need multiple ISSNs. But we all know that 'should' doesn't always happen (even in the best link resolver, there are gaps and errors in the knowledge base--and there can be errors in the generated CO too, maybe one of the ISSNs the generator knew about was in fact in error), and what's wrong with redundancy? I think redundancy in this situation is good, gives the software more chances to recover from errant or missing metadata (which is a fact of life) and still provide a succesful outcome. Seems to me, if sending multiple ISSNs (or ISBNs) would harm link resolver performance, then _that's_ something to blame on a link resolver that's not smart enough. If the generator knows lots of things about the work cited (including multiple ISSNs for th! e serial that holds it), why not send all the information on for the link resolver to use however it may (or may not) choose? Jonathan Eric Hellman [EMAIL PROTECTED] 02/28/08 12:51 PM I agree; issn is not an identifier for an article. But in general, a resolver should be smart enough to know what serial is meant even if a variant issn is supplied. I do not agree that it would be helpful for generators to send multiple issn's. They currently can send issn and eissn; if the resolvers knowledgebase is good, then sending it multiple issns will never help and will often degrade its performance. I think its sucky to craft OpenURL metadata that caters to substandard resolvers. On Feb 19, 2008, at 11:09 AM, Jonathan Rochkind wrote: Be careful though, please don't send an rft_id ISSN identifier for an _article level_ metadata package. OCLC does this. It's wrong, as the ISSN does not serve as an identifeir for the _article_ cited, but rather for the journal it's in. Until I figured out what was going on, this caused some bugs in Umlaut. It _would_ be nice to send multiple ISSNs even for an article-level citation. Let's say the generator of the OpenURL happens to know that there are several variant ISSNs for the publication, that identify differnet manifestations, but any of which are valid for the given article citation. It would be helpful for the generator to send them all along, in case the link resolver knows about some but not all of them, to increase the chances that the link resolver will properly 'recognize' the citation. But, it unfortunately can't be done. It's not the end of the world to realize that OpenURL isn't perfect (what is? By trying you learn from your experience to do better next time), but I'm unconvinced that this is actually desirable in any way instead of an oversight. One thing I think I feel like we've learned from many of our community's recent metadata initiatives is the importance of creating standards in such a way that they can be further developed and/or extended in a backwards compatible way. Ie, an OpenURL 1.1 or something, that was backwards comptable so it could be sent to resolvers that knew no more than 1.0 without problems. This has to do with both the design of the structure/syntax of the metadata, as well as the design of the _processes_ of maintenance, to make this kind of extension and development not too cumbersome socially. Jonathan Karen Coyle wrote: Ah, you're referring to rft_id, and I was looking at the ISBN element in the KEV Book format. So using rft_id would work. The reason for multiple ISBNs is that many MARC records have ISBNs for the hard copy and the paperback. Without going through some gyrations you don't know which is which, although for purposes like ILL either is valid. There are also multi-volume works that each get an ISBN. Like other FRBR levels manifestation has a fairly wide range of ambiguity. A book simultaneously published in two countries... is that one manifestation or two? What if they each get a separate ISBN? A hardback and trade paperback that come out at the same time, where the only difference is the cover... and the ISBN? Although I often use the shortcut of ISBN = manifestation the fact of it is that ISBNs are publisher inventory and sales numbers and are used in ways that are convenient for publishers. They also get mis- assigned frequently, as some tests being run on bib data at the Open Library are showing. kc Ross Singer wrote: Actually, this: http://alcme.oclc.org/openurl/servlet/OAIHandler/extension?verb=GetMetadatametadataPrefix=mtxidentifier=info:ofi/fmt:kev:mtx:ctx indicates that multiple rft_ids *are* valid, and, in fact, would have to be, since you could very easily have a DOI and a PMID and, say, a SICI. I have no idea what any resolver would do with this bundle of ISBNs, of course. It also seems somewhat contrary to the intention of the Book metadata
Re: [CODE4LIB] Multiple ISBNs in COInS?
Be careful though, please don't send an rft_id ISSN identifier for an _article level_ metadata package. OCLC does this. It's wrong, as the ISSN does not serve as an identifeir for the _article_ cited, but rather for the journal it's in. Until I figured out what was going on, this caused some bugs in Umlaut. It _would_ be nice to send multiple ISSNs even for an article-level citation. Let's say the generator of the OpenURL happens to know that there are several variant ISSNs for the publication, that identify differnet manifestations, but any of which are valid for the given article citation. It would be helpful for the generator to send them all along, in case the link resolver knows about some but not all of them, to increase the chances that the link resolver will properly 'recognize' the citation. But, it unfortunately can't be done. It's not the end of the world to realize that OpenURL isn't perfect (what is? By trying you learn from your experience to do better next time), but I'm unconvinced that this is actually desirable in any way instead of an oversight. One thing I think I feel like we've learned from many of our community's recent metadata initiatives is the importance of creating standards in such a way that they can be further developed and/or extended in a backwards compatible way. Ie, an OpenURL 1.1 or something, that was backwards comptable so it could be sent to resolvers that knew no more than 1.0 without problems. This has to do with both the design of the structure/syntax of the metadata, as well as the design of the _processes_ of maintenance, to make this kind of extension and development not too cumbersome socially. Jonathan Karen Coyle wrote: Ah, you're referring to rft_id, and I was looking at the ISBN element in the KEV Book format. So using rft_id would work. The reason for multiple ISBNs is that many MARC records have ISBNs for the hard copy and the paperback. Without going through some gyrations you don't know which is which, although for purposes like ILL either is valid. There are also multi-volume works that each get an ISBN. Like other FRBR levels manifestation has a fairly wide range of ambiguity. A book simultaneously published in two countries... is that one manifestation or two? What if they each get a separate ISBN? A hardback and trade paperback that come out at the same time, where the only difference is the cover... and the ISBN? Although I often use the shortcut of ISBN = manifestation the fact of it is that ISBNs are publisher inventory and sales numbers and are used in ways that are convenient for publishers. They also get mis-assigned frequently, as some tests being run on bib data at the Open Library are showing. kc Ross Singer wrote: Actually, this: http://alcme.oclc.org/openurl/servlet/OAIHandler/extension?verb=GetMetadatametadataPrefix=mtxidentifier=info:ofi/fmt:kev:mtx:ctx indicates that multiple rft_ids *are* valid, and, in fact, would have to be, since you could very easily have a DOI and a PMID and, say, a SICI. I have no idea what any resolver would do with this bundle of ISBNs, of course. It also seems somewhat contrary to the intention of the Book metadata format, since I think it's (in my murky view of FRBR-y terms) trying to define a manifestation rather than the expression level that Bill is trying to use it for. I could be weaving in my own interpretations and biases there. An alternative would be use by-reference context objects and then make the context objects available as XML. You could have multiple context object available in one XML document this way. A combination of COinS/unAPI could make something like this possible. -Ross. On Feb 18, 2008 6:53 PM, Karen Coyle [EMAIL PROTECTED] wrote: Actually, the max occurrence of ALL of the KEV keys is 1 except for au (which is unlimited). I remember discussions in which we acknowledged that one key NE one value, eg you could input multiple values if your recipients were in agreements (a poor excuse, I know). Thus: isbn:;isbn:. My only memory for why max=1 for all of these is that it has to do with the fact that there is no structure or dependency in KEV, so an OpenURL with keys rft.au=nnnrft.title=tttrft.au=rft.title= isn't interpretable in terms of what authors go with what titles. Why the exception for au but for no other fields? My memory fails me here. Undoubtedly it made sense at the time. kc Jay Luker wrote: Hi William, According to the book KEV format (defined here: ttp://tinyurl.com/2psmkq) the max occurrence of the isbn key is 1. I'm assuming that by extension that means that the rft.m-key (i.e., rft.isbn) form is also limited to one occurrence. So specifying multiple ISBNs that way is a no go. You can however specify multiple referent identifiers. From the KEV Context Object format matrix (http://tinyurl.com/2r5hsc): Multiple instances of rft_id do not indicate multiple Referents, but rather multiple ways to identify a single Referent So
Re: [CODE4LIB] Multiple ISBNs in COInS?
Hi William, According to the book KEV format (defined here: ttp://tinyurl.com/2psmkq) the max occurrence of the isbn key is 1. I'm assuming that by extension that means that the rft.m-key (i.e., rft.isbn) form is also limited to one occurrence. So specifying multiple ISBNs that way is a no go. You can however specify multiple referent identifiers. From the KEV Context Object format matrix (http://tinyurl.com/2r5hsc): Multiple instances of rft_id do not indicate multiple Referents, but rather multiple ways to identify a single Referent So I *think* what you could do is this: rft_id=urn:isbn:isbn1rft_id=urn:isbn:isbn2... Also, I'd be remiss not to point you to a more authoritative list for OpenURL questions: http://listserv.oclc.org/scripts/wa.exe?A0=OPENURL. Although I'm sure there's plenty of overlap in interest/knowledge in the subject between the lists. -- Jay Luker [EMAIL PROTECTED] Software Engineer, Ex Libris Inc. (617) 332-8800, x604 http://www.exlibrisgroup.com On Feb 17, 2008 3:14 AM, William Denton [EMAIL PROTECTED] wrote: I'm hep to the COInS scene now and am using it in some lists of books I'm generating. For some of the books I know multiple ISBNs. Can I include them all in one COInS span somehow? Doing one individually makes my OpenURL Referrer extension clutter up the page with a lot of links. I looked at the specification but it didn't seem to cover this. generator.ocoins.info only seems to want one one ISBN. Putting multiple rft.isbn variables just makes the last one overpower the earlier ones. Any tips appreciated! Bill -- William Denton, Toronto : www.miskatonic.org www.frbr.org www.openfrbr.org
Re: [CODE4LIB] Multiple ISBNs in COInS?
I'm not sure if you can do that in the KEV format that OpenURL uses, although you could do it in the XML format. But it still woudln't mean exactly what William wants it to mean--and most existant link resolvers wouldn't neccesarily do the right thing with it. OpenURL can be a bear sometimes. I would also encourage you to send that question to the OpenURL listserv. Jonathan Jay Luker wrote: Hi William, According to the book KEV format (defined here: ttp://tinyurl.com/2psmkq) the max occurrence of the isbn key is 1. I'm assuming that by extension that means that the rft.m-key (i.e., rft.isbn) form is also limited to one occurrence. So specifying multiple ISBNs that way is a no go. You can however specify multiple referent identifiers. From the KEV Context Object format matrix (http://tinyurl.com/2r5hsc): Multiple instances of rft_id do not indicate multiple Referents, but rather multiple ways to identify a single Referent So I *think* what you could do is this: rft_id=urn:isbn:isbn1rft_id=urn:isbn:isbn2... Also, I'd be remiss not to point you to a more authoritative list for OpenURL questions: http://listserv.oclc.org/scripts/wa.exe?A0=OPENURL. Although I'm sure there's plenty of overlap in interest/knowledge in the subject between the lists. -- Jay Luker [EMAIL PROTECTED] Software Engineer, Ex Libris Inc. (617) 332-8800, x604 http://www.exlibrisgroup.com On Feb 17, 2008 3:14 AM, William Denton [EMAIL PROTECTED] wrote: I'm hep to the COInS scene now and am using it in some lists of books I'm generating. For some of the books I know multiple ISBNs. Can I include them all in one COInS span somehow? Doing one individually makes my OpenURL Referrer extension clutter up the page with a lot of links. I looked at the specification but it didn't seem to cover this. generator.ocoins.info only seems to want one one ISBN. Putting multiple rft.isbn variables just makes the last one overpower the earlier ones. Any tips appreciated! Bill -- William Denton, Toronto : www.miskatonic.org www.frbr.org www.openfrbr.org -- Jonathan Rochkind Digital Services Software Engineer The Sheridan Libraries Johns Hopkins University 410.516.8886 rochkind (at) jhu.edu
Re: [CODE4LIB] Multiple ISBNs in COInS?
Actually, the max occurrence of ALL of the KEV keys is 1 except for au (which is unlimited). I remember discussions in which we acknowledged that one key NE one value, eg you could input multiple values if your recipients were in agreements (a poor excuse, I know). Thus: isbn:;isbn:. My only memory for why max=1 for all of these is that it has to do with the fact that there is no structure or dependency in KEV, so an OpenURL with keys rft.au=nnnrft.title=tttrft.au=rft.title= isn't interpretable in terms of what authors go with what titles. Why the exception for au but for no other fields? My memory fails me here. Undoubtedly it made sense at the time. kc Jay Luker wrote: Hi William, According to the book KEV format (defined here: ttp://tinyurl.com/2psmkq) the max occurrence of the isbn key is 1. I'm assuming that by extension that means that the rft.m-key (i.e., rft.isbn) form is also limited to one occurrence. So specifying multiple ISBNs that way is a no go. You can however specify multiple referent identifiers. From the KEV Context Object format matrix (http://tinyurl.com/2r5hsc): Multiple instances of rft_id do not indicate multiple Referents, but rather multiple ways to identify a single Referent So I *think* what you could do is this: rft_id=urn:isbn:isbn1rft_id=urn:isbn:isbn2... Also, I'd be remiss not to point you to a more authoritative list for OpenURL questions: http://listserv.oclc.org/scripts/wa.exe?A0=OPENURL. Although I'm sure there's plenty of overlap in interest/knowledge in the subject between the lists. -- Jay Luker [EMAIL PROTECTED] Software Engineer, Ex Libris Inc. (617) 332-8800, x604 http://www.exlibrisgroup.com On Feb 17, 2008 3:14 AM, William Denton [EMAIL PROTECTED] wrote: I'm hep to the COInS scene now and am using it in some lists of books I'm generating. For some of the books I know multiple ISBNs. Can I include them all in one COInS span somehow? Doing one individually makes my OpenURL Referrer extension clutter up the page with a lot of links. I looked at the specification but it didn't seem to cover this. generator.ocoins.info only seems to want one one ISBN. Putting multiple rft.isbn variables just makes the last one overpower the earlier ones. Any tips appreciated! Bill -- William Denton, Toronto : www.miskatonic.org www.frbr.org www.openfrbr.org -- --- Karen Coyle / Digital Library Consultant [EMAIL PROTECTED] http://www.kcoyle.net ph.: 510-540-7596 skype: kcoylenet fx.: 510-848-3913 mo.: 510-435-8234
[CODE4LIB] Multiple ISBNs in COInS?
I'm hep to the COInS scene now and am using it in some lists of books I'm generating. For some of the books I know multiple ISBNs. Can I include them all in one COInS span somehow? Doing one individually makes my OpenURL Referrer extension clutter up the page with a lot of links. I looked at the specification but it didn't seem to cover this. generator.ocoins.info only seems to want one one ISBN. Putting multiple rft.isbn variables just makes the last one overpower the earlier ones. Any tips appreciated! Bill -- William Denton, Toronto : www.miskatonic.org www.frbr.org www.openfrbr.org