Metadata

I was talking with J tonight, in particular regarding data storage schemas. In particular freeform writing and its associated metadata.

If everything were mostly-freeform attribute text, it would be a search-friendly index-hating chaotic mess.

On the simple, gentle end of things, attribute values with spelling errors are a problem. One that can be easily mitigated with on-the-fly ‘did you mean X?‘ prompts.

On a the end of chaos, attribute names with random text in them could render any attempts to make sense of the data, and provide a browsing interface, futile at best.

However, keeping a firm, rigid list of fields that can be filled in is, in a best-case scenario, somewhat limiting. With many fields, this could become very inefficient very quickly. Thus, the search for a happy medium leaves one with the need for a strict, but extensive list of optional metadata. Dublin Core provides the sort of MetaData elements that would be required, in a handy-dandy standardized set.

The possibility of intristically supporting multiple instances of each attribute is appealing. eg: Filk, Kimiko, Tragic

About this entry

You’re currently reading “Metadata”, an entry on VerseLogic

Published:: 09.06.04 / 12am

Category:: Uncategorized

AboutMe

Alan J Castonguay is from Ontario, Canada. Interested in philosophy, animation, serial art, and mixing bleeding-edge technology with doing things the old fashioned way. Works in tech support, but dreams of going back to the theatre.

A resume is available for your perusal, should you be interested in bribing me to do work.

_Quinn 09.21.04 / 12am

You do get a bunch of metadata for free; time and date is a big one. Title(s) are free (or almost so); even an untitled piece could profitably be referred to by its first line. One aspect of metadata you don’t normally think about in computing but most hand-writing authors use when flipping through archives is _shape_. The combination of shape and date can be really powerful for browsing [through large search result sets]: I can dig up the reference for you later, but displaying timelines/calendars of documents with thumbnails of them (their first page) has proven to be very effective for finding documents. (You may want to make them selectable for a ’search selected documents’ search widget.)

One idea I’d like to look into more is deriving metadata from searches and search results. Something along the lines of ‘an unsuccesful search on “Kimiko” followed by a succesful search on “brightly cute” might mean that Insert Title Here might have something to do with Kimiko.

VerseLogic

Metadata

About this entry

1 Comment

Have your say

AboutMe

RecentlyWrit

Theme Switcher