Thank you, Nathan. This is very much what I'm trying to accomplish, and incorporates a lot of the elements that have gone into my design. It appears to mimic the actual process that my tool is designed to implement. I particularly like that it takes advantage of the wiki's cross linking ability to allow the users to dig deeper into a more complex issue. As an overall tool, though, it is still locked into a wiki model, which results in certain inherent limitations. Your comment here illustrates this pretty well.
On 4/3/06, Nathan Spaeth loplin@gmail.com wrote:
- Throughout the process (through a method yet unkown to me, maybevoting?) every assertion would have a score that decided how much it actually supported one side or the other, if it adequatly countered any other assertions, etc.
For each assertion to be voted on, this would require that some metadata be maintained about that assertion. How many votes, direction of votes, who has already voted and in what direction so that they might change their vote later, that kind of thing. Wikis are largely self moderated, so this metadata would need to be tracked and stored by each of the participants or by a central authority.
By stepping out of the wiki model we can embody each of the assertions as its own data object. Assertion, creator, votes, links, the whole nine yards, stored as a living object in the database. This solves the metadata problem and allows you to automate the entire process. My design correlates to Wikibates the same way that dynamic pages correlate to static ones.
I also suggest checking out OpenCyc (http://www.opencyc.org). It is a
knowledge based AI with all kinds of assertions about real world things. This might be usable as a backend when forming arguments. You could make assertions to the AI when adding to the debate, the AI could then use them to check the validity of other assertions made in the current or even other debates. I definately thinking rating system should be added so that specific arguments can carry more weight over others.
I'll have to look deeper into OpenCyc. A quick glance suggests that it could be a very powerful back end resource, and may even provide a good data model to work with.
Thanks again,
Robert