An Australian startup believes that the best way to protect children online is through an artificially intelligent software program.
The inventors of this idea are banking on children's attachment to pets. The creature's 'cuteness' and helpfulness will ingratiate the software with the child, so that he or she will respect it and listen to it, or even find it as a likable companion. Agent-based internet applications are nothing new but for concerned parents, this might be an admirable solution to what is perceived by many to be a growing problem. From one of the inventors: 'Of course, we're also planning to release a version of the Moji IM for teenagers and adults, but we're focusing on children at the moment.'"
star-techcentral.com/tech/story.asp
The highly complex areas of artificial intelligence and computational linguistics seem to allow it to learn from the user and react in real-time.
It does so by using natural language processing techniques to analyse or parse text in a user's speech at a semantic level.
It derives keywords from the speech and tries to match them to the user's interests stored in its database using statistics and probability formulas.
How would the ambiguity of semantics and syntax involved in language ever be overcome with this type of AI software?
The inventors of this idea are banking on children's attachment to pets. The creature's 'cuteness' and helpfulness will ingratiate the software with the child, so that he or she will respect it and listen to it, or even find it as a likable companion. Agent-based internet applications are nothing new but for concerned parents, this might be an admirable solution to what is perceived by many to be a growing problem. From one of the inventors: 'Of course, we're also planning to release a version of the Moji IM for teenagers and adults, but we're focusing on children at the moment.'"
star-techcentral.com/tech/story.asp
The highly complex areas of artificial intelligence and computational linguistics seem to allow it to learn from the user and react in real-time.
It does so by using natural language processing techniques to analyse or parse text in a user's speech at a semantic level.
It derives keywords from the speech and tries to match them to the user's interests stored in its database using statistics and probability formulas.
How would the ambiguity of semantics and syntax involved in language ever be overcome with this type of AI software?
posted by:
|
SF Bay Area
|
-
Re: Artificially Intelligent Software - Mor(f)
Tue, January 23, 2007 - 11:09 PMIts not doing real natural language parsing. Sounds like they are using a "bag of words" statistical model to map from detected words to categories. This is similar to how automatic document classification systems work, and the technique is used by most major search engines as well. It kind of side-steps the semantic issue by disregarding semantics entirely and concentrating only on which words appear in context with each other.
