jump to navigation

Resource Description is by humans March 24, 2008

Posted by Mia in FRBR, cataloguing.
Tags: , ,
add a comment

RDA – Resource Description and Access. That is, how humans might go about describing resources (books, journals, etc.) in a consistent fashion. RDA’s raison d’etre is to be read and applied by humans. In Coyle’s presentation slides from code4lib 2008, I was intrigued by this declaration on slide 14:

Another problem is that RDA instructs its adherents to create text strings.

It does? Ok, I guess I missed it. Contentious issues notwithstanding, RDA is a set of description guidelines with examples. I haven’t been pouring over it or following the RDA list, etc., but the example shown on this slide follow a rule stating: “Record the extent of a resource… [blah blah].” If humans have to do the recording, well, they probably have to be shown an example, since these help us conceptualize.

So for me, what the examples do is supply some prototypes to help me understand what is meant by the phrase “extent of cartographic material“. The example “6 maps on a sheet” is a conceptualization, which I then apply in some other framework for subsequent machine processing.

Would “Capture the extent of a resource…” be a less offending choice of terminology? Or “tag”? “Markup”? etc.

The point of humans being able to apply the description guidelines is to make the act of describing the resource something which is understandable and recognizable. Also teachable, but moreso, readily learnable. Ideally, the examples would be so transparent that upon seeing the prototypical example, one can readily apply to many situations henceforth.

To look at it differently, one could interpret “Record the extent…” to mean instead: “Turn on your microphone [recording device], verbalize the following concepts in spoken language, then turn off the device.”

Recognizing things is fundamental to conceptualizing.

Chronological facet in FAST March 8, 2008

Posted by Mia in FRAD, FRBR, Uncategorized.
Tags: , ,
add a comment

One of the criticisms of FAST is that the chronological facet is separated from the topic. So if there is a span of years, the facet becomes meaningless since it is devoid of context. Let’s take the following LCSH heading:

Furniture|z United States|x History |y19th century

and convert it to FAST. Since each facet can only be subdivided by a facet of the same kind, i.e., topic subdivides by topic — this heading converts to 3 FAST headings:

Topical: Furniture — History
Geographical: United States
Chronological: 1900-1999

The chronological facet — and the geographical as well — sits by itself. When there are multiple headings with geographical and/or chronological, these are split from the context and by themselves are at best meaningless, and at worst, the dots can be connected to the wrong topic.

So something is missing –and that is the relationship of these terms to the other terms. While this is a legitimate criticism of FAST at this stage, it should be up to the system’s retrieval and presentation logic to do the appropriate combining. Why would we expect any user to intuit this or any other construction?

A user would probably not begin a search for “1900-1999″ in and of itself. A user will first search for a something (or a someone), so typically, a noun. If the user were interested in the loose concept “Nineteenth century” that by itself could well be the main topic, rather than an aspect of something else, say the concept Art, or Literary Criticism, modified by a ‘1900-1999′ facet. So, is there any correlation, say an inverted order between the occurrences of the term and its importance in the search string, in determining the “predominance” of the facet? And if so, how would a system compensate?

The musings continue.