Re: [RDA-L] libraries, society and RDA
Three years or so ago I thought, finally the significance of what authority data can do for improving data management is understood but more recently it seems to have been lost in the dust. I would add to Bernhard and Jim's comments that the rules governing the construction of authority data for automated management are long overdue. Too much of the data and rules are designed for human intervention. So much of the focus in current discussions is on bibliographic records rather than authority records, which is really backwards. A name authority record should have as much data as possible on a person. Ideally, all known works should be added to a name record, with additions over time. The relationship to the work should be provided (author, director, actor, performer, etc.). Birth and death dates, and other identifying information should all be provided in a manner to help identify other outside resources like online biography sources. Mary L. Mastraccio Cataloging Authorities Librarian MARCIVE, Inc. San Antonio Texas 78265 1-800-531-7678 [EMAIL PROTECTED] Thanks for highlighting this Mary. I hope the following makes sense (and is correct!) I think this is absolutely correct, and chimes extremely well with the work that Karen, Diane et al. have been doing with RDA - the realisation is that what is true for Author's or Subject's is true for other (possibly less complex) aspects of metadata. Even with something as simple as 'physical description', can benefit from a separation between the metadata record and the detail of the physical description. For example, if we consider 300$b (other physical details), which maps to a number of RDA elements, including 'Production Method' (http://www.collectionscanada.gc.ca/jsc/docs/5rda-parta-ch3rev.pdf, p.26 in doc not as numbered). Karen, Diane et al. have created a relevant vocabulary at http://metadataregistry.org/vocabulary/show/id/33.html - click on 'Concepts' tab to see the list of values. So, rather than inserting the 'literal' value lithograph into 300$b, these vocabularies open up the possibility of linking to http://RDVocab.info/termLIst/RDAproductionMethod/1007 - which is the URI for the term lithograph. RDA refers to this as a 'non-literal value surrogate' - this comes from the language of Dublin Core Metadata Initiative, but basically means that you point at the value using a URI, rather than the use of the literal string 'lithograph' (or whatever) as we do with MARC. As with using Authority files for Author's etc. this use of separated vocabularies opens up the possibility of saying more about lithographs - what they are, alternative terms, translations etc. A weakness of MARC21 is that it doesn't make use of the reference to the Authority record into the metadata record - we rely on 'literals' too much - making it more difficult to ensure consistency, make changes, or draw into our indexing information not held directly in the MARC record. I have to admit that I my understanding of RDA gets a bit hazy at this point, but from what I can see it also treats elements such as 'Creator' as non-literal value surrogates - and anywhere this is true, we can treat as a possible point to establish a link to an Authority file. Owen
Re: [RDA-L] libraries, society and RDA
Owen Stephens wrote: The question of 'feasibility' takes us beyond a question of whether it is 'worth it' to whether it can be done. What the report says is that the authors do not believe it is possible to achieve consistency with metadata of judgement except within a tightly controlled, narrow and consistent environment - and the repository environment is not any of these things. This is not just about cost, but about people and their behaviour. No argument here. I was being generous in my original message, but I went on to say that it is possible to achieve consistency in metadata of judgement. I will go on to say that people have relied precisely on this consistency for over a hundred years, if not far longer, and for someone to say that it isn't feasible is an unjustified conclusion, in my opinion. Certainly, this consistency is not 100%, and people must be trained to do it correctly (I fear that current training in subject analysis and heading assignment is not improving). For many years, studies have shown that two different, highly-trained people will assign different subjects to the same item. My reply is: so what? This ignores the power of the syndetic structure of the catalog, where users can find related terms and therefore find everything. Perhaps one cataloger assigns Despotism while another assigns Authoritarianism, users can still use the syndetic structure to find the works. Humans may not hit the bull's-eye each time, but they will come close, and with the use of the structures, things should be found. Compare this to computer systems automatically assigning terms that are completely off the mark. Instead of either of the headings above, a computer may come up with Military art and science or x-ray photography. I realize that general understanding of the use and importance of the syndetic structure is not appreciated, and this is probably because it is so poorly implemented in our current catalogs. Before concluding that something that has been relied upon for such a long time is not feasible, a little more work should take place and the alternatives need to be explored in depth. I will be the first to agree that deep and profound changes are needed and that automated subject assignment is improving and may actually work someday. But not today. Jim Weinheimer
Re: [RDA-L] libraries, society and RDA
Jim, I recognise that you disagree with the report (and me!) that consistency in metadata of judgement is feasible. I would say that whatever consistency has been achieved in the past (and I'm not convinced it is as consistent as you suggest here - if it was, then issues such as FRBRisation, deduping etc. would not be half the problem they are today) has been achieved by trying to create a 'controlled environment' with partial success - I believe that the advent of the web fundamentally changed our ability to control our environment. However, all this aside (which we could I'm sure discuss for sometime preferably over a glass or several of something nice), what I think we should be clear on is that the report under discussion does not make an argument of automatic creation of metadata of judgement - such as subject headings. The report says explicitly about deep/broad records relating to metadata of judgement that computer technology is not yet at a stage to replace human effort in this regard. The report also is doubtful that automatic creation of 'brief' records of metadata of judgement is possible at the current time saying brief judgemental records are the domain of humans (and maybe computers) Owen Owen Stephens Assistant Director: eStrategy and Information Resources Central Library Imperial College London South Kensington Campus London SW7 2AZ t: +44 (0)20 7594 8829 e: [EMAIL PROTECTED] From: Resource Description and Access / Resource Description and Access [mailto:[EMAIL PROTECTED] On Behalf Of Weinheimer Jim Sent: 14 November 2008 12:56 To: RDA-L@INFOSERV.NLC-BNC.CA Subject: Re: [RDA-L] libraries, society and RDA Owen Stephens wrote: The question of 'feasibility' takes us beyond a question of whether it is 'worth it' to whether it can be done. What the report says is that the authors do not believe it is possible to achieve consistency with metadata of judgement except within a tightly controlled, narrow and consistent environment - and the repository environment is not any of these things. This is not just about cost, but about people and their behaviour. No argument here. I was being generous in my original message, but I went on to say that it is possible to achieve consistency in metadata of judgement. I will go on to say that people have relied precisely on this consistency for over a hundred years, if not far longer, and for someone to say that it isn't feasible is an unjustified conclusion, in my opinion. Certainly, this consistency is not 100%, and people must be trained to do it correctly (I fear that current training in subject analysis and heading assignment is not improving). For many years, studies have shown that two different, highly-trained people will assign different subjects to the same item. My reply is: so what? This ignores the power of the syndetic structure of the catalog, where users can find related terms and therefore find everything. Perhaps one cataloger assigns Despotism while another assigns Authoritarianism, users can still use the syndetic structure to find the works. Humans may not hit the bull's-eye each time, but they will come close, and with the use of the structures, things should be found. Compare this to computer systems automatically assigning terms that are completely off the mark. Instead of either of the headings above, a computer may come up with Military art and science or x-ray photography. I realize that general understanding of the use and importance of the syndetic structure is not appreciated, and this is probably because it is so poorly implemented in our current catalogs. Before concluding that something that has been relied upon for such a long time is not feasible, a little more work should take place and the alternatives need to be explored in depth. I will be the first to agree that deep and profound changes are needed and that automated subject assignment is improving and may actually work someday. But not today. Jim Weinheimer
Re: [RDA-L] libraries, society and RDA
Mary Mastraccio wrote: I think the child/title authority record is needed rather than just using the works information to allow for name/title headings in bib record displays. It might be determined that the Works field in the parent record should only have the control number of the child name record. [I use control number but certainly a URI could be supplied.] I strongly suspect that the various language translations of a title could/should really be in the one NameTitle/Child record. To make this work the title display would probably relate to a language code in the bib record and the NameTitle/Child record could have multiple Title fields. I better stop with this or the email will get too long. This is the type of work I would like see happen. All that's needed in MARC for this to happen was suggested a long time ago, and submitted as a discussion paper (meanwhile accepted by MARBI): http://www.loc.gov/marc/marbi/2007/2007-dp01.html Subfield $0 is for the link. Links are necessary always _only_ from the lower level to the upper level, not both ways! Or think what it would mean otherwise: you would have to enter a link twice (at both ends) and update the upper level record (e.g. for a series) every time a new lower level record arrives (e.g. for a volume). Proper indexing can take care of correct linking. We've been doing these things in our (non-MARC) environments for decades. The Pica software comes complete with all necessary functions. It is owned by OCLC. An older description is here: http://www.allegro-c.de/formate/reusep.htm B.Eversberg
Re: [RDA-L] libraries, society and RDA
A weakness of MARC21 is that it doesn't make use of the reference to the Authority record into the metadata record - we rely on 'literals' too much - making it more difficult to ensure consistency, make changes, or draw into our indexing information not held directly in the MARC record. MM: No doubt new technologies and formats could allow us to do better data management more easily, however, I find arguments that include the phrase weakness of MARC21 usually look at the current AACR2 rules or implementation of MARC21 rather than MARC21 itself. There are many things that could currently be done in MARC if 1. systems were designed to use the data, 2. the data was consistently entered, 3. [not required but would make life easier] some modifications should be made to the MARC fields to allow the same type of data to be entered all in one place at one time. For example: Catalogers now have several places in the fixed fields to enter form/format information and then again in 245$h (not to mention other fields with subfield $h), possibly 300, 6xx subfield $v and then in field 655. This is what causes inconsistency in data input and a major reason why systems are not designed to harvest that data. This problem isn't with MARC, so the same issues are important when designing and implementing a new format. I think that in this case I am talking about a weakness in MARC21 rather than AACR2. Possibly it is the implementation - but isn't MARC21 an implementation of MARC? As an example, I cannot see how MARC could support a linked data approach to information stored in fixed fields - one of the places where it would benefit from it. However, I do agree that new technologies are formats could allow us to do better data management more easily! I also believe that even if it is possible to 'tweak' MARC to allow more use of non-literals this wouldn't necessarily be easier than designing a new format. Moving on to your example of authority records - this looks fine to me, but is this covering the ground that FRBR and FRAD have started on - which in turn RDA is building on? Owen
[RDA-L] Language of text strings in RDA
It occurs to me that one thing that might be useful in future versions of RDA would be a provision for routinely making explicit the language of text strings, in order to facilitate language-dependent services (such as did you mean? suggestions). While the language of cataloging (MARC21 040$b) may be a fairly accurate predictor of the language of notes and other constructed text strings, and the language of the item (MARC 21 008/35-37 and 041) may be a fairly accurate predictor of other text strings, both are probabilistic with regard to any given string. In RDA, the language of cataloging (language of the description) is a contextual given rather than something that is explicitly recorded, and the instructions relating to recording the language of the content support only a context of human interpretation and not machine manipulation. I think in future it will be much more useful to have the language of the string-or the contextual language of the string in cases where it includes the odd substring in a different language or script-made explicit by an associated language code (the default value of which would be the language of the item). Anyway, just a thought. Ed Jones Assistant Director, Assessment and Metadata Services National University Library 9393 Lightwave Avenue San Diego, California 92123-1447 +1 858 541 7920 (voice) +1 858 541 7997 (fax)
Re: [RDA-L] libraries, society and RDA
Owen Stephen: I think that in this case I am talking about a weakness in MARC21 rather than AACR2. Possibly it is the implementation - but isn't MARC21 an implementation of MARC? MM: At one level MARC21 is an implementation of MARC but what I am thinking of is the library systems implementation of MARC (use of the MARC data) and obviously to some level how MARBI decides to store AACR2/RDA data in MARC21. Many of our problems are related to how the rules tell us to put the data in MARC, not with the format itself. OS:As an example, I cannot see how MARC could support a linked data approach to information stored in fixed fields - one of the places where it would benefit from it. MM: Systems already use MARC fixed fields information to display icons related to format and to refine searches and there are drop down boxes with additional information. There is no reason a system design cannot use the code in the fixed fields to provide additional information to a user. OS:However, I do agree that new technologies are formats could allow us to do better data management more easily! I also believe that even if it is possible to 'tweak' MARC to allow more use of non-literals this wouldn't necessarily be easier than designing a new format. MM: I agree with you that the amount of work tweaking MARC could be significant so would be better spent on designing a new format, but unless the designers recognize that a major problem with MARC is design--not the format--they are not going to properly address those design issues. OS: Moving on to your example of authority records - this looks fine to me, but is this covering the ground that FRBR and FRAD have started on - which in turn RDA is building on? This IS covering the ground that FRBR and FRAD claim to be covering but I and many others are growing frustrated in not seeing any progress beyond the theorizing that began years ago and virtually nothing on FRAD (where some of think we need to start). Progress might be happening in secret somewhere but those of us waiting eagerly in the wings are beginning to think it isn't going to happen, or if it does, it will not be as well designed as it could be. I do think RDA is trying to build on the FRBR/FRAD effort but their work might be easier if they had a clearer structure around which to build their documentation. Yes, I know that RDA is supposed to free us from structure, but we all know there has to be some structure or they can reduce RDA to put in what you want. Mary L. Mastraccio Cataloging Authorities Librarian MARCIVE, Inc. San Antonio Texas 78265 1-800-531-7678 [EMAIL PROTECTED]
Re: [RDA-L] libraries, society and RDA
OS:As an example, I cannot see how MARC could support a linked data approach to information stored in fixed fields - one of the places where it would benefit from it. MM: Systems already use MARC fixed fields information to display icons related to format and to refine searches and there are drop down boxes with additional information. There is no reason a system design cannot use the code in the fixed fields to provide additional information to a user. This is slightly different - I agree systems can be intelligent about how they use coded information, but I very strongly believe that we need to use linked data, not codes that need system interpretation. We may be in danger of violently agreeing on the basic premise here :) This IS covering the ground that FRBR and FRAD claim to be covering but I and many others are growing frustrated in not seeing any progress beyond the theorizing that began years ago and virtually nothing on FRAD (where some of think we need to start). Progress might be happening in secret somewhere but those of us waiting eagerly in the wings are beginning to think it isn't going to happen, or if it does, it will not be as well designed as it could be. I do think RDA is trying to build on the FRBR/FRAD effort but their work might be easier if they had a clearer structure around which to build their documentation. Yes, I know that RDA is supposed to free us from structure, but we all know there has to be some structure or they can reduce RDA to put in what you want. I suspect that many would agree - it is frustratingly slow. However, as we can see from discussion on this list, there is still much work to be done to persuade people that (a) change is needed and (b) RDA is a step in vaguely the right direction. As you probably know, the Working Group on the Future of Bibliographic Control seem to recommend that more fundamental work needed to be done to 'test' FRBR, and work on RDA should be suspended until this was done: 3.2.5.1 JSC: Suspend further new development work on RDA until a) the use and business cases for moving to RDA have been satisfactorily articulated, b) the presumed benefits of RDA have been convincingly demonstrated, and c) more, large-scale, comprehensive testing of FRBR as it relates to proposed provisions of RDA has been carried out against real cataloging data, and the results of those tests have been analyzed (see 4.2.1 below) However, LoC together with NAL and NLM responded by saying that: Until the completion of the rules and the availability of the RDA online tool, reviewers will not be able fully to assess their impact on: Description, access, and navigation practices for a broad array of users and types of materials Current and future electronic carriers and information management systems to support RDA goals Estimated costs for implementation and maintenance during a time of flat, even reduced, budgets Essentially saying that RDA should be completed, and then an assessment carried out - now it has been announced that there will be a delay to the 'online tool' I'm not sure whether this will change the statement above (i.e. will an assessment of impact be possible with only the text available?) It feels like a real chicken and egg scenario - we are unlikely to see substantial effort going into systems supporting RDA unless it is adopted by the community. The community is unlikely to adopt RDA if there are no systems that support it... Owen
Re: [RDA-L] libraries, society and RDA
Diane: I actually think that OCLC hasdone a very goodjobintegrating and making copy-specific information accessible on the Web. We wereworried about making the transition from RLIN, with its individual records for each institution's copy of an item, to WorldCat, with its master record structure. We're quite happy with the way our institutional records work well and play well with master records. Now if they could just adjust the displays for visual materials so that they worked the same as for books or manuscripts, all would be gas andgaiters. Liz O'Keefe "Diane I. Hillmann" [EMAIL PROTECTED] 11/13/2008 11:13 PM Adam:Thanks for pointing this out--I've had little experience using WorldCat,and wasn't really aware of these particular problems. This seems socounterintuitive--libraries have been complaining for years that theirvendors have made so little use of the richness of MARC records, andhere's OCLC building systems that use even less. Liz's additional pointabout how the data is "dumbed-down" then reconstructed ignoring datathat specifies the relationship of the person to the resource suggeststo me that OCLC might be going the way of Ford/GM/Chrysler. The currentdownward trajectory of those three companies suggest that there isn'tmuch hope for organizations who stick with their broken strategiesdespite all indications that the world is changing around them.But at least the big three aren't requiring those who own their cars towear teeshirts with their logos everywhere they go.DianeAdam L. Schiff wrote: Yet still, if you insist to be interested in none else than "Miller, Henry - 1891-1980", then as of yet there are only library catalogs or WorldCat to turn to. This ought to change, no? B.Eversberg Actually, you need to strike mention of WorldCat above, as it makes no use of authorized forms of names, and doesn't even show them. A search in WorldCat for Henry Miller does A KEYWORD SEARCH on the words Henry and Miller. You are never presented with a list of Henry Millers from which you could narrow your search. In the "Refine Your Search" window, under "Author", you do get Henry Miller as a facet, and the results are indeed pretty good, but they will also include works by and about the other Henry Millers in WorldCat records. For example: Study of the at-risk population of infants and toddlers in Maryland : final report by Henry Miller; Nancy Peterson; Maryland. Infants and Toddlers' Program Interagency Coordinating Council.; Center for Health Policy Studies (U.S.) 1989 The principles and practice of obstetrics. Including the treatment of chronic inflammation of the uterus, considered as a frequent cause of abortion. by Henry Miller 1858 The saucy little widow down in Pimlico by Henry Miller 1864 In displays such as above, the designers of WorldCat decided (god knows why!) to remove any and all additions to names such as fuller forms of name and dates. This means that users who wonder if the well-known author Henry Miller is the same as the medical writer or composer of The saucy little widow are left to figure that out some other way. Those statements of responsibility in the WorldCat displays, by the way, are not taken from the statements of responsibility actually found in the OCLC record, but are made up by taking all the main and added entries, reversing the inverted forms, and throwing them all together preceded by the word "by". Adam Schiff University of Washington Libraries [EMAIL PROTECTED]
Re: [RDA-L] libraries, society and RDA
On Thursday, November 13, 2008 10:13 PM, Diane I. Hillmann wrote: This seems so counterintuitive--libraries have been complaining for years that their vendors have made so little use of the richness of MARC records, and here's OCLC building systems that use even less. Unless something has changed recently, it seems to be the trend in several of the more recent projects. For example, VuFind, Koha, Evergreen, at least by default or in their demo configurations, use the 1xx heading rather than 245$c statement of responsibility, like Worldcat.org as described by Adam Schiff. Also lacking in many of the new projects is the ability to do alphabetical heading browses like those possible at http://authorities.loc.gov/ and http://catalog.loc.gov/. Bryan Baldus Cataloger Quality Books Inc. The Best of America's Independent Presses 1-800-323-4241x402 [EMAIL PROTECTED]
[RDA-L] RDA a step in the right direction?
Owen Stephens said: ... there is still much work to be done to persuade people that (a) change is needed I know no one who does not think change is needed. and (b) RDA is a step in vaguely the right direction. I know of no person whose opinion I have come to trust, who agrees RDA is a step in the right direction. Apart from its poor writing style, fragmentation of descriptions, and optional elements (as opposed to required if relevant), it addresses the wrong target. What needs targeting is, (1) as Michael Gorman suggests, simplification of AACR2 (with special material provisions removed to supplementary codes, and more standardization (place of publication jurisdiction provision, treaty entry, etc.); (2) as Mary of Marcive suggests, the creation of standards for authority records allowing linking of related manifestation records; (3) as many have suggested, developing ILS software which utilizes the wealth of AACR2/MARC21 records (will Evergreen/Pine/Sitka achieve this?); and (4) to obtain the interoperability so many wish, the aggressive promotion of ISBD among all producers of bibliographic metadata. __ __ J. McRee (Mac) Elrod ([EMAIL PROTECTED]) {__ | / Special Libraries Cataloguing HTTP://www.slc.bc.ca/ ___} |__ \__