W3C home > Mailing lists > Public > public-sws-ig@w3.org > October 2005

LAM technologies for SW

From: Linda Ridgeway <linda@alierra.com>
Date: Fri, 7 Oct 2005 23:20:27 +0300
Message-ID: <03af01c5cb7c$923b7300$0905a8c0@201vitaliy>
To: <public-sws-ig@w3.org>


When surfing the web, I found something that got my attention and I'm really curious about that. That was the use of LAM (Linear Associative Memory) for filtering information. As far as I know, associative memory is any neural network that deals with associative learning and retrieval of information. It remembers patterns of data it has been given, and one retrieves those patterns by giving it similar and/or partial data that have been associated with the different patterns. I didn't know it's possible to create such an architecture, based on neuropsychological principles: That's what really amazes me! Of course, if we go deeper into the issue, the ability to predict user's behavior doesn't seem mysterious at all. User activity is stored in a database like an individual's experience is stored in his head (either in consciousness  or subconscious), and exact recall of data when given only partial input is possible on the Internet as well as with a real man. Please correct me if I'm wrong: I'm wondering if such technologies are used in SW Research & Development. I believe that could be really useful for SW science: However, I have some questions that I can't answer myself. If we use LAM for SW, do we use just user activity information or can we use LAM in Ontologies,  i.e. associations based on the meanings of words? Everybody knows, human mind is complicated, and it's hard to predict the behavior of a human even in real life. How can we do that in computer Science? Is that possible at all?


This topic is very interesting for me, and would be happy to discuss all that with those who know more than I do. 


I hope some of W3.org visitors share my interest and look forward to getting some opinions!




Linda Ridgeway
Marketing and Research Manager
Received on Friday, 7 October 2005 20:20:46 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:54:15 UTC