[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
Re: navigation ideas
- To: <xtech@xxxxxxxxxxx>
- Subject: Re: navigation ideas
- From: Paul Baclaski <acad!megalon!peb>
- Date: Mon, 30 Apr 90 17:33:11 PDT
- Cc: <megalon!rick>
How about "found paths" instead of Subscription? Similar to "found
sound," a term used in music for using sounds from the environment in
a composition. Subscription seems too overloaded with meaning and
it is a minor synonym of "endorse". The idea with found paths is
that the paths are more or less discovered by the reader: look for
the well worn paths in the forest and avoid the dense bushes.
This term does not cover the case where a single published path by a
known author is used. In this case, you are a tourist following a
tour guide. (The terms endorsement or subscription make sense
in this case.) I have no idea of a single term that encompasses
both the tour guide and found path ideas.
As you mention, the consensual tour includes directionality like
a Markov chain (n-order?). Adding many peoples ideas of a good
path together might end up with a garbage: I can foresee having
a simple neural net (summation, weights and threshold) to compose
a prospective path. The user can slide a slider around representing
the threshold and view the %nodes visited (as well as the number of
nodes to be visited); then, when a suitable % is found, the path is
accepted and can be saved/bookmarked, etc. I am assuming that loops
are automatically removed early in this process. I like to use the
term Internerary for this: it is a locally stored object (FE) that
the user is treking through on and off over the course of days. As
a person who reads many books at a time, I'd like many itineraries
that I can save and restore. An itinerary could become stale, but
as a reader, I should have a choice of creating new ones and using the
"when visited" information on my old itinerary to mark off stuff I've
already viewed on the new itinerary (and disposing of the old one).
The simple neural net could be composed by the sophisticated user
or one could be selected from a set of generic ones (e.g., Pure
Markovian, or These Readers/Authors, or combination). The simple
net could be displayed with sliders at each threshhold. The composed
net is like a DAG or could be implemented as a simple expression
An inverse operator would be useful for finding out what no one (or few)
has ever read or no one cares about. This would help authors
(hmm, no one notices/sees that link...) and arcanephiles.
Afterthought: Have two completely separate mechanisms: found paths
and Tours. The tours are known authors and the found paths are
completely anonymous. Usually, one will create an intererary out
of one or the other. However, a converter is used to translate a
set of tours into weighted raw data for composing a pool of raw
chains out of which the itinerary is created.
This may be entirely identical to the neural net idea, especially if
each tour can be converted with a separate weight into the raw chains.
However, it might be easier to explain to the person who thinks
neural networks are complex, and thus be a simpler user model.
P.S.: Another term for "found paths" could simply be "consensual tours".