Difference between revisions of "Talk:Bibliographic/OOoBib Functional Requirements/Keywords"

Revision as of 14:34, 26 September 2006

comments from Bruce

I actually don't think this is a good idea. It involves far too much complication for minimal benefit. The same can be achieved using keywords/tags. Add some auto-completion for them, and problem largely solved.

Note: I'm not saying there isn't room for controlled vacabularies, where they are in effect, full resources/objects. Indeed, my own RDF collection works just this way. It's just that this sounds like a very complicated thing to implement in what needs, I think, to be a fairly stripped down component. Also, it will not be possible to come up with any standard scheme that everyone (or even a majority) will agree on.

comments from Oliver

I'm a scientist in life sciences (forest-ecophysiology to be exact) and I have been playing with reference software for the past 10 years. I have also played with keyword, standardized thesauruses, have tried to analyse/adapt them etc. Generally I don't find them useful. Following some comments on the differnet subjects on the wiki page : Journal categories might be a good idea for some people who use these specialised kind of journals, however all of the Journals I use would either be "Basic Research" or a mix of different categories Keywords : I'm very much in favor of hierarchical keywords and aliases. The hierarchy is important so that when you search for example articles on trees, you not only search for all articles with the keyword "tree", but also those with "beech", "oak", "pine" etc, . Aliases are important so that you don't end up with synonym keywords and the example on the wiki page with the insurace company illustrates this nicely. However, where it all becomes very very complicated is when you try to standardize this across social, earth and life sciences. Try putting three scientists together, even from close fields (let's say life science, a geneticist, a physiologist and an ecologist) and have them agree on the definition of some concept (keyword). Won't work. And this is not because they are stubborn, but simply because they have different points of view and requirements. I have looked at "professional" hierarchical keyword thesauruses (such as used by CAPS abstracts or Current Conents) and I have tried to adapt them for my personal use. This did not work at all and this is because they have tried to accomodate all kinds of different sciences. This is also why often people say that they don't find what they are looking for when searching literature. They just need to have a look at the thesaurus their database is using to realize that the keywords they were looking for don't meanat all what they thought they would mean. This is why I think a predefined hierarchy would not be of any use and actually a lot of lost effort that could be put elsewhere. You can try to create a manageable thesaurus for a small workgroup, but actually creating the thesaurus is not only putting up a list but actual scientific work, defining the relationships between different concepts. However effort could be put into assuring that the creation and management of a "personal" thesaurus is as easy as possible. A number of tools spring to mind such as:

• attaching notes to keywords to be able to precisely define the keyword. This also helps others to understand your keyword (I actually now use a wiki to organize my comments on articles I'm interrested in and I only use the reference database to format citations, actually the ideal solution in my opinion would be the possibility to link the keywords in a reference database to a wiki system)
• automatic search/verification of doubles eg in aliases
• automatic keyword detection (of existing keywords) in title and abstracts and the possibility to chose which of those should be used
• comparison of different thesauruses and importation of parts of thesauruses
• etc etc

Journal List : I don't think that creating a journal list on this wiki page is of any use : I have made some statistics and the dozen or so colleagues in my work group, all in my specialized field, look regularly at some 300 different journals. Journal lists with abbreviations and access urls can be found easily on the web. Oliver

Reply

I don't think that creating a journal list on this wiki page is of any use.

Indeed, this is much work, but some journals need the full name for a citation, others need the abbreviated name (and still others may need the very short abbreviation, as I recently learned). It would be ugly to have to write by hand for every reference both Journal names. This should be indeed done automatically (and therefore the need for this list).

Tools

Indeed, some of the tools you have mentioned would be quite useful.

• automatic keyword detection: I thought of something similar (using standard keywords), but of course, localising the keywords is a good alternative to standard keywords

Journal categories

• this is quite important if one is creating a huge database for a university or big department (instead of many local DBs for every user, which I believe should be the ultimate goal)
• Journal Categories would be here of great importance
• most Journals belong to a given category, and there isn't any need for great (or any) user input; therefore, once the Journal list is complete, there is not much work to do anymore

Reply Oliver on Journal List

there is no need to establich a list of journal abbreviations here as most of the large Libraries do have one. See for example http://www.nal.usda.gov/catalog/download_jia.shtml where you can actually download the list as access file. And if you don't find your journals in there, there are others....

Oliver

Hierarchical Keyword Tree

I have expanded the Hierarchical Keyword Tree Section on the article page. I believe that this is one of the most important concepts that emerged here and I would appreciate any comments and further examples on this topic. I created a medical example, but it would be nice to see this working for non-medical exmples, too.