To: Snowshoe who wrote (789 ) 11/25/2002 2:06:10 PM From: E. Charters Read Replies (1) | Respond to Good site, thanks. Yes google trees are amazing. The world is linked to the next greatest or next least thing most efficiently through google nodes. It is amazing, isn't it, that in a binary search tree the furthest an item can be away is the root of all the things in the heap to the base 2. So in a 100 billion words, a tree search for the next greater or lesser thing is no more than 39.86 node interrogations or steps. So *much* for the math of data structures. Google may use tries or other dictionary forms of indexing though. Trie alpha-paths can be even shorter and collisions are not a problem in related searching. Optimization of searching with or for more than one thing is the not-usual part of string searching in this case. Today you could keep a respectable index of the entire internet's pages, their uber-relations, their subjects and subtopics on two 80 gige drives for 500 bucks worth of media. One hundred million pages would index minutely in 150 giges, not even compressed. Someone should do that and sell the data and crawler service to keep it updated. Fit on 5 compressed double sided DVD's. Could sell specialty info for less and much smaller space perhaps 10 gige of "biopharm" on 50 index keys, with a CD, with crawler service too. Has to benefit business. Have to do that soon as the search engines will go out of business soon. World Com was their death knell. Advantage of service would be higher relevancy, faster sub searches for relevancy, data mining, and overnite relevancy searches and compilations automated.
