Saturday, September 17, 2005

Leobard had some comments about SemanticBible (but really about New Testament Names):

I wish that they would have identified people not by name but by identity and name.

there are several homonyms in the new and old testament, would be great to match bible verses to people and not names.

I'm not multilingual enough to figure out how to leave a comment on his blog, but this represents both an important point and an important misunderstanding. One thing that moves NTN beyond current word-based search interfaces is exactly this distinction: when you look for "Mary", you want to distinguish Jesus' mother from Mary Magdalene, from Clopas' wife Mary who seems to be mother of James and Joseph. NTN does make this distinction: it's one reason i'm still saying there are about 350 instances of Man, because i'm still teasing them all apart, and so i don't actually know yet how many there are.

Once i complete the basic knowledge base, the next step will be to attach a list of verse references to each, so you can know for a given verse which Mary it refers to, and vice verse. I expect to do this automatically, though the homonyms will likely require a little manual disambiguation.

It's better when constructing RDF identifiers (not just names) for people to give them some resemblance to names: "MaryWifeOfClopas" is more user-friendly than"genid:ARP2147041". This raises the question of how to do this consistently, since as far as i know i'm inventing standards here. The principles i've used, stated informally, are the following:

  • The most basic name forms should be used for the most common names: so "Mary" is the mother of Jesus, "John" and "Simon" are Jesus' disciples.
  • Traditional or conventional names are used whenever possible (e.g. JohntheBaptist), with the usual camelcase convention. For example, the four Herod's mentioned in the New Testament are "HerodAgrippaI", "HerodAgrippaII", "HerodAntipas", and "HerodTheGreat".
  • Names from the two genealogies of Jesus (Matt.1.1-16, Luke.3.23-28) which often have little additional information, are disambiguated when necessary by including their parent's name (so "Alphaeus" is distinct from "AlphaeusFatherOfLevi", and "CainanSonOfEnosh" is distinct from "CainanSonOfArphaxad").
  • Functional descriptions are used for other distinctions: "AlexanderTheCoppersmith" is distinguished from "AlexanderOfTheSanhedrin" and "AlexanderTheHeretic".
  • If no other distinguishers are available, native place is used, e.g. "SimonOfCyrene", "AlexanderOfEphesus".

I'm not sure whether anybody else in the world cares about this somewhat obscure issue, but i'd love to engage in a dialog with others who have relevant ideas.

12:02:13 PM #  Click here to send an email to the editor of this weblog.  comment []  trackback []

Blogged here to help me find it later is an interesting article by Scott Berkun (who worked on the UI for Internet Explorer, and has recently switched to Firefox) about how to build a better browser. This is relevant to me on two fronts:

  • i'm interested in building usable interfaces for searching Christian information, both the Bible and other references like New Testament Names that i hope will become useful guides
  • most of my professional life is spent on developing technology to help people find things, particularly in the growing pile of unstructured information available on the Web. Though there are important differences between professional researchers (like government analysts) and casual surfers, many principles are the same

I'm particularly interested in following up on his overview of how people use browsers, and why bookmarks are so dumb. His two examples of tracking richer meta-data and frequency of visitation seem obviously useful.

10:13:37 AM #  Click here to send an email to the editor of this weblog.  comment []  trackback []