[libdb-develop] FRBRish/LibDB Chat Transcript

[Morbus I. <mo...@di...>]

Impromptu discussion of FRBR/LibDB on irc.freenode.net.
When I'm there, I'm always in #disobey and #swhack. The
unedited form is archived permanently at the following URI:
http://notabug.com/swhack/chatlogs/2004-02-20.html#T15-27-44

---

  <deltab> you noted that users would differ in how strictly
  they'd apply rules, how much they'd learn of ontologies, etc.

  <Morbus> correct, yes.

  <deltab> it might be a good idea to store what class of user enters data

  <Morbus> so that then upper levels would be able to say "show
  me data from people who know what they're doing"?

  <Morbus> deltab: so, you want that lvl applied to rows, not the whole db?
  rows would be the only way to say "show me something to fix".

  <deltab> yes

---

  <deltab> did you notice the section in FRBR about parts?

  <Morbus> i've noticed the section on parts, but I don't remember
  it 100%. my impression on the first read through was that it was
  related to individual physical parts of a whole: the study guide
  to an english book, the troika in a music CD, etc.  the parts that
  have been discussed on the list have been related to parts of
  a singular item. pg 3 through 7, minutes 6:43 to 6:59, etc.
  i didn't think the frbr parts dialog covered that stuff. but,
  again, i don't recall specifics. i've done no specific coding
  or development work on "parts" as a concept.

  <deltab> me neither. I wanted to read that section again

  <Morbus> i wanted to read a bunch of stuff about movie encoding
  too before i did any "real" thoughtwork on it. er, not movie
  encoding per se, but movie markup related stuff. i've a document
  somewhere that talks about movie metadata.

  <deltab> you're not daunted by the potential massive scale of
  it? so many people, so many pieces of work assembled

  <Morbus> not really. i know what i want, and i strive to get it.
  well, yeah, i mean, the amount of reading to be done sucks.
  especially considering that no one can agree on anything. but,
  i'm equally positive that whatever choice i make, a bunch of
  people are gonna say "you should use my idea instead".

---

  <deltab> have you decided exactly what will be described and what won't?

  <Morbus> ultimately, its up to the end-user, but i've modeled
  one of my own movies in the mysql_sample.sql the movie in the sample
  schema models production companies, hair and makeup assistances,
  and defines objects on a lievel like "swimwear". it also keeps track
  of the characters played by a person, and the artisan (distribution
  company) catalog number. i certainly don't expect "casual" users
  to care about that stuff. and i have no doubt that librarians aren't
  gonna give a crap about the "artisan catalog number".

  <Morbus> how much you put in is really up to you. but the data that
  i've been building around as my base assumes a high level of detail.
  personally, for mem, i'm gonna be marking up how people die in a
  movie. so, my personal LibDB will include a "Death By" annotation.
  which will be searched/indexed/shown along with all the other
  annotations, etc. you'll be able to show all "death by garden tools"
  in "films" of the "1980s", for instance.

  <deltab> I suppose even something like Darth Maul's contact
  lenses could be considered works,

  <Morbus> well, the user could do that if they wanted. i'm certainly
  not going to stop them. but it doesn't make any goddamn sense.

  <deltab> it might for the designer :-)

  <Morbus> it doesn't fit well in the frbr model, and it doesn't
  have any true identifiers (unless you consider retail model
  number duplicates, patent numbers, etc.) but, yeah, if you wanted
  to make a contact lense database, sorted by color (concepts),
  objects (stars, spirals, cats eye), or event (halloween, swimming),
  you could do that.

  <deltab> and have them connected to the films in which they're used?

  <Morbus> deltab: yes, because you'd be able to define a)
  relationship types ("ie. film prop") and b) relationship
  ("this WORK is a FILMPROP of WORK")

  <deltab> so it could get arbitrarily complex
  nice in theory, but could chew up lots of time

  <Morbus> yes. the user determines how complex. [chewing up
  lots of time] doesn't sound any different from any other
  metadata-galore system though. i think the only people who
  would truly *use* *use* the metadata system would be the
  second level of user. the first (casual) user wouldn't want
  to put that much effort in. the third (librarian) wouldn't
  have TIME to put that much effort in. so only the second
  (discriminating) user would attempt to. but, that's not to
  say that it won't be possible for the other users.

---

  <sbp> so it'd be nice if you could search within the
  labels that you put in. will you allow that? so all deaths
  that contain the word "garden", etc.

  <Morbus> sbp; yes, that'd be possible. annotations can be searched
  by. i don't have any plans (yet) to make a topic map interface,
  but the db supports it. a concept can have a relationship to
  another concept. so, the "Rake" object could be a RT to the
  "Garden Tool" object. you'd then be able to browse objects,
  concepts, places, etc., in a hierarchal sense. or, show related
  terms on searches, in listings, etc.

---

Thoughts?


-- 
Kevin Hemenway