 |
 |
Hi Sarah Perhaps "didn't understand your posting" might be more accurate. But perhaps also the use of "keyword" as an element name in the Classification section is the source of confusion. My understanding of it was that the Classification:Keyword element was where you put the terms selected from the taxonomy specified in the TaxonPath. So perhaps "Classifier" or some such might have been a better element name than "Keyword". Putting my point in your terms, I was trying to suggest that maybe (only) classification should be done in the Classification section and keywording should be done using the General:Keyword element. As to the different types of users, I was simply recalling some of the discussions that took place when the original IMS Metadata spec was being drawn up way back in '98-'99 as it helped resolve some of the disputes that were taking place then (the content publishers who tacitly assumed all metadata would be embedded in the content, and therefore that there would be one definitive MD record per LO as created by them, and keywords selected by them would suffice, vs cataloguers who tacitly assumed all metadata was independent of the content, held in separate catalogue/repository/index systems, and therefore that there could be multiple records per LO, and support needed for proper classification with taxonomies and their source). This seemed to have some echoes in the discussion here and it maybe was part of the reason for it being incompletely resolved in the original IMS MD and hence in IEEE LOM. I am not sure that I agree with points 5 & 6 (if I understand them correctly ;-). Having made the distinction between classification and keywording, you then seem to suggest that it is up to whoever creating the metadata to decide on which to use and how, depending on their situation. But searchers have to know how an element is being used in order to frame an effective query. It therefore seems, at a minimum, that this is a question to be determined by a community of use rather than by 'whoever', and if the community wants clarity on the semantics and use (and hence interoperability) then that agreement is best set out in an application profile, but it would be better still if it was clear and unambiguous in the standard. If, going the other way, you feel that a standard should be flexible and allow different people to do things in different ways, it should still set out the alternatives as MAYs where it doesn't set them as MUSTs or SHOULDs. This approach then leaves it up to communities of use to fix on their preferred approach in an application profile. But specs, standards and application profiles are fundamentally about agreements on how we work together and these then inevitably act as agreed constraints. Areas that are left to the individual to decide (e.g. the text that goes into a field) should have no impact on interoperability. But this seems to me to be an area where the interpretation and use of elements will have an impact on interoperability. So if LOM and the UK LOM Core are to allow for all the options you give, then they need to be explicitly supported in the spec/standard/ap, maybe by adding an optional 'VocabPath' element under General 1.5:Keyword to support your option 4 (if, as you say, there is a need for it), with examples showing how they are used. If it's not clear in the LOM/UK LOM Core documentation how keywording and classification are supported, then I think you have highlighted a point of ambiguity, and hence of potential interoperability breakdown, which should be clarified in a future release of LOM/UK LOM Core. Cheers, Bill Sarah Currier wrote: > Hi Bill, > > I get the feeling you didn't actually read my email! > > The scenario you mention is likely a valid use case under certain > circumstances. However that is really up to whoever is making the > decisions about the metadata management for that repository or project > or whatever to decide. > > My point is that: > 1. Classification and keyword indexing are two different things with > different fucntions. > 2. Both can be done by anyone, be they publisher, resource author, > cataloguer, or whoever. > 3. Keyword indexing can be done with or without a controlled vocabulary. > 4. It may be useful to be able to declare what controlled vocabulary > is being used when one is being used (but I'm not sure about this one > and have yet to hear an actual practical argument either way). > 5. Decisions about how the above four things are done within a given > situation should be up to whoever is making the decisions for that > situation given what their needs, context and circumstances are. > 6. The LOM and the UK LOM Core should therefore allow for all the > above options (including 4. if the wider community shows there is a > need for it) and not dictate which they think people SHOULD be doing. > > Cheers > Sarah > > Bill Olivier wrote: > >> Hi Sarah >> >> I wonder if the questions you ask arise from MD serving two different >> but related roles/purposes: >> 1. The Publisher's description of their LO (e.g. the publisher's front >> page of a book (is there a name for this?)) >> 2. Independent Librarians/Catalogers' classification of a LO (e.g. a >> catalogue card) >> >> Not surprisingly, I think you are wearing the second hat. >> >> I would think that LO publishers should use General 1.5:Keyword when >> they want to add whatever terms they think relevant, especially if they >> are not familiar with available taxonomies. >> >> My understanding (but its a long time since I looked at it!) is that the >> classification section's 9.4:Keywords should only contain the relevant >> terms from the vocabulary/taxonomy specified in the 9.2:Taxonpath. >> >> So if you have a known vocabulary use the 9:Classification section, if >> not use 1.5:Keyword (a place for non-librarians/ taxonomists to play >> :-) . >> >> My understanding is (please correct me if I'm wrong!) that you are >> permitted to have multiple instances of 9:Classification and therefore >> this is the place to use multiple taxonomies when needed. e.g. in the >> context of Grainne's development of three taxonomies for Learning >> Activities (Context, Pedagogical Approach, Task), each could have a >> separate classification element as part of a profile using LOM for >> Learning Activities (which will doubtless cause all sorts of debate >> about whether a learning activity can be a learning object...) >> >> Bill >> >> Sarah Currier wrote: >> >>> Hi Scott >>> >>> Thanks for teasing this out a bit more for us non-techies! This does >>> look useful. >>> >>> Scott Wilson wrote: >>> >>>> If something is being used as a keyword rather than a classification, >>>> I would suggest the source of the keyword isn't that critical, >>> >>> >>> >>> >>> I need to think about this. You may well be right... interesting. >>> Might consult some of my Info Science colleagues across the road. >>> >>> Ahh, the CETIS Metadata SIG. It's so kewl isn't it, getting answers, >>> having assumptions challenged, etc. etc.? >>> >>> Sarah >>> >> >> >