Steve Richfield said:
Does anyone else here share my dream of a worldwide AI with all of the
knowledge of the human race to support it - built with EXISTING Wikipediaand
Dr. Eliza software and a little glue to hold it all together?
Hi Steve,
I share part of your dream, in that I am strongly attracted to Wikipedia as
great corpus of commonsense knowledge that should be incorporated into my AGI
project, Texai. I've looked at the Freebase Wikipedia Extraction:
The Freebase Wikipedia Extraction (WEX) is a processed dump of the English
language Wikipedia. The wiki markup for each article is transformed into
machine-readable XML, and common relational features such as templates,
infoboxes, categories, article sections, and redirects are extracted in tabular
form.
Freebase WEX is provided as a set of database tables in TSV format for
PostgreSQL, along with tables providing mappings between Wikipedia articles and
Freebase topics, and corresponding Freebase Types.
My plan:
1. Rather than deal with Wikimedia, collaborating on a AGI-style
interface , I would simply process (i.e. parse and completely understand) their
content.
2. One could then teach Texai how to edit a Wikipedia article to close
the loop.FYI. Another great corpus of knowledge is an online patent database.
In the US, patents include a section that describes the background of the
invention. On the road to generally applicable machine vision, I can envision
a facility that can read various types of patent diagrams.
Cheers,
-Steve
Stephen L. Reed
Artificial Intelligence Researcher
http://texai.org/blog
http://texai.org
3008 Oak Crest Ave.
Austin, Texas, USA 78704
512.791.7860
----- Original Message ----
From: Steve Richfield <[EMAIL PROTECTED]>
To: [email protected]
Sent: Friday, May 16, 2008 9:31:58 AM
Subject: [agi] AGI and Wiki, was Understanding a sick puppy
Mike, Stan, et al,
I have recently had some interesting off-line discussions that may be pulling
things together, so I thought that I would run the emerging concept up the
flagpole here and get any opinions.
I have previously posted here the horrific problems trying to deal with the
Wiki people, as they apparently perceive value in impeding a good AI/AGI
interface to Wikipedia. Apparently, even their own internal people can't get
help here, as it might lead to loosing control over the search business.
However...
The Wikipedia software is open source, and some companies even maintain their
own domain-specific wikis as knowledge base. Further, the AI/AGI interface
problem is certainly not their only problem. The 2nd biggest problem is their
implicit insistence on a single model/paradigm behind every article, which
limits Wiki to being of value only for grade-school support.
Note in passing the value of faulty models. Often the most accurate model
suggests no means of correction, whereas a less accurate model suggests
corrections that quite often (but sometimes don't) work (see puppy update
below). Limiting articles to single models destroys MOST of the potential value
of Wikipedia, as does blocking an AI/AGI interface.
Proposal: Start a new AI/AGI Wikipedia, starting with the present open-source
Wikipedia software with minor mods to collect additional information from
authors and build a database on an associated FTP site for anyone to download.
This should soon take over the Wiki business from the present Wikipedia folks.
A well placed patent application would impede their following suit, thereby
seizing this entire marketplace.
Unfortunately, this is too big of a project to be funded with my lunch money or
built and maintained with my limited spare time. However, with an investor to
cover miscellaneous expenses, a server to hold the site, and some
co-conspirators to help make it go; and this could quite easily take over
much/most of the Internet in a way that would be MUCH bigger than ever
envisioned by Wikipedia.
Does anyone else here share my dream of a worldwide AI with all of the
knowledge of the human race to support it - built with EXISTING Wikipedia and
Dr. Eliza software and a little glue to hold it all together?
Note that unlike the present Internet, that Dr. Eliza is pretty much
language-independent. You can even put in a problem statement in one language,
and get the unanswered questions and analysis out in another language. The
principles underlying this are similar to financial systems that keep the
numbers in a database, and use different language versions of their program to
access it, only in Dr. Eliza, nearly every record has a field to indicate
language so that no software changes are needed to support different languages,
though trivial enhancements ARE needed to support new languages with previously
unsupported features, e.g. the differing use of periods and commas in numbers
depending on which side of the pond that you reside on. With this, the WHOLE
world would be automatically included, rather than just the English language
part of it (with trivial separate participation by other languages) as is
presently the case. No longer would the Internet be
divided up according to languages.
Puppy Update:
The puppy is doing MUCH better, and is now starting to explore. Three new
theories as to its problems have emerged:
1. Pus found on its fur pointed the way to an abscess in its armpit that had
evaded previous inspection. The abscess seems to be too small to be
life-threatening, but who knows?
2. It has an umbilical hernia that might be strangling some intestines.
3. Like some people, it would apparently rather die than eat puppy food,
though it doesn't seem to be picky about eating minced leftovers.
However, earlier theories, though probably incorrect, DID guide the way to
treatment that, though not perfect (it would have been nice to lance the
abscess), was sufficiently successful to save its life. This serves to
highlight the value of incorrect theories, that they often provide the right
answers in a timely manner, even when for the wrong reasons.
Steve Richfield
________________________________
agi | Archives | Modify Your Subscription
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=103754539-40ed26
Powered by Listbox: http://www.listbox.com
