Ai Dreams Forum
Member's Experiments & Projects => AI Programming => Topic started by: FuzzieDice on January 21, 2007, 06:00:54 am
-
MIT changed things around now and old links to OMCSRAW is changed. You can download it here:
http://commonsense.media.mit.edu/Resources.htm
The problem is, this is all the sentences but in a gzip file (Winzip will open it). Furthermore, all the lines are preceeded by a number now, and not just plain lines of sentences. So you'll have to write a program to parse out the numbers before each sentence on each line.
Other than that, it's 20MB of data compressed into a ~377KB compressed gzip file.
Hope it might be useful to those still wanting to use OMCSRAW in their projects.
-
great find, thanks for the link.
-
I think they changed it again, this link was working the other day, but today it isn't. I was womdering how useable it would be in an AIML bot.
-
How bizarre, here today gone tomorrow. I guess it's get it while the gettings good at MIT
-
Yeah, seems so..if anyone comes across this again, please post it up again ::)
-
http://commonsense.media.mit.edu/cgi-bin/download.cgi
-
Thanks Dan :)
-
rats, it wasn't a good zip
-
The zip goes 404.
I just checked here: http://csc.media.mit.edu/Resources.htm
Again. And at first it didn't work. Now it is. :/
-
Yay! Thanks Fuzzie, now to parse it.
-
Don't tempt me, Dan. LOL!
Seriously I wish I could write a program to do that but right now I'm swamped with stuff.
-
You or someone else can always fork the database snapshot into "unedited" ... I think omcsraw.txt is a little big for one wiki pageThe best she best source is a file from Mit made for bots,omcsraw but it is 48 Mb. ... as the information is stored and retrieved from a database the QUERYS are SQL
-
Bruce,
Welcome to our humble site!!
Just an FYI, the post you've responded to, while helpful, was / is almost 3 years old.
Fuzziedice is no longer an active member here.
Omscraw was severely dated when the original posting was made but in the realm
of AI, every resource is potentially, a good resource.