Wiki has a summary.
I am going to do some searching and theorizing and maybe some proving regarding the link flow in a semantic net. I will start that we organize our information searching so as to minimize the flow from general to specific. So let's set up a set of link factories, network nodes from which there is clicking through. So we would expect pages to organize such that they participate in the optimum fashpn, as part ot the optimal flow.
We can use an aggregate sniffer to hunt down patterns were the flow look like all the -iLog(i) are within an integer. I will work on how to find those maximum entropy information channels, and in the process smoke out the big search vendors.
Wiki, for example, sets out to become the maximum entropy semantic network, they work hard on optimizing the descent from general to specific. In doing so, Wiki maximizing the construction of an idea from, say, a multi-semantic factory. They pre-organize the search graph and their clients get trained on the optimum search construction. So the search has dimensions, the rank of the match patterns. The solutions, the 'eigen vectors' are going to be finite sets of words that form the canonical inventory in a graph match. I will try to create the wiki semantic forms by search. In doing so, spook the search vendors, always fun.
No comments:
Post a Comment