[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
Re: Fuzzy Hypertext?
- To: xanadu@xxxxxxxxxxxxx
- Subject: Re: Fuzzy Hypertext?
- From: gtn@xxxxxxx (Gavin Nicol)
- Date: Fri, 20 Dec 1996 16:54:08 -0500
- In-reply-to: <Pine.GSO.3.93.961219115754.11261B-100000@uhunix4> (message from Art Pollard on Thu, 19 Dec 1996 12:02:26 -1000)
- Reply-to: xanadu@xxxxxxxxxxxxxxxxx
- Sender: xanni@xxxxxxxxxxxxxxxxx
>> When an author creates a link, I would like them to be able to give the
>> link a fuzzy level, so that they can express uncertainty eg links which are
>> definite, sometimes useful, only useful to the dedicated reader
>Perhaps this is best implemented by some sort of user modeling.
>Otherwise, it could be very hard for a hypertext author to decide that
>"this text is only 50% relevent" while "this text is 74% relevent."
This is precisely why I thought associative memory might be an
interesting thing to use... and in particular, as implemented in a
NN. In such systems, the memory "learns" by strengthening connections
based on "relevance" (ie. do these two nodes fire at the same time).
In simple implementations of such systems they require N*N
connections, where N is the number of possible link positions (start
or end of links). In practise, you can have such systems "grow"
connections, and have connections "die".
An interesting feature of such a system would be that you could filter
the number of returned connections by specifying a threshold, possibly
also using a maximum retrieval time period.
Such a system would learn, and forget links, and learn which are more
"relevant" as traversal patterns change.