Mark Waser >>Could you educate us some? Not really. Many SQL problems have solutions/workarounds, but we let cost-benefit analysis decide. We found that for most of our requirements our own solutions were more effective than using SQL. We did invest several man-months on these issues.
Peter -----Original Message----- From: Mark Waser [mailto:[EMAIL PROTECTED] Sent: Wednesday, February 21, 2007 2:03 PM To: [email protected] Subject: Re: [agi] Development Environments for AI > PS. Regarding Databases, in our own work we use SqlServer for some things > (it's an integral part of our overall system), but found it quite useless > for central knowledge representation. My personal WAG (wild-assed guess) is that you can/want to use SqlServer for two purposes: a) similar to long-term memory and b) context/swap space. I certainly wouldn't want to use it for working memory (which seemed more of your thrust than what I would call central knowledge representation). > There were many specific > problems we encountered in trying to make more use of a SQL. Could you educate us some? ----- Original Message ----- From: "Peter Voss" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Wednesday, February 21, 2007 4:05 PM Subject: [agi] Development Environments for AI >A few comments on this debate: It's good to keep in mind the different > premises that people hold -- > > There are those who believe that no-one knows how to build AGI at this > stage. For them, discussing reverse-engineering the brain or building new > experimental/exploratory development environments may make sense. > > The there are those who believe that current hardware has inappropriate > architecture and/or technology, or is *way* underpowered (many orders of > magnitude). Not much hope there, unless you believe that you can design & > build such new systems AND you know what AGI will require! > > Lastly, there are those of us who believe that there are indeed some > people > who know how to build AGI *now*. Hopefully, those would concentrate on > finding the best approaches, and making it happen ASAP. > > Peter Voss > http://adaptiveai.com/ > > > PS. Regarding Databases, in our own work we use SqlServer for some things > (it's an integral part of our overall system), but found it quite useless > for central knowledge representation. The three main reasons are: many > real-time inserts (they are slow), the need for very large numbers of > simple > queries that depend on previous results (single query overhead kills > performance), and the need for specialized data requirements (very sparse > tables, highly dynamic table creation, etc.). There were many specific > problems we encountered in trying to make more use of a SQL. > > > > > ----- > This list is sponsored by AGIRI: http://www.agiri.org/email > To unsubscribe or change your options, please go to: > http://v2.listbox.com/member/?list_id=303 > ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?list_id=303 ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?list_id=303
