Saturday, November 12, 2011

Add schema links to G World

Why not? A schema is a link property, not a link function.  So, simply invent an operator, choose an entirely confusing punctuation mark for it and add it to typeface Explosion.  There is no violation to function down only 'hard' links, as long as the residual link function is derived, specifically, the sibling/descending order.  In other words, go hunting for schema links, and make more of them all you want.  Pick the : operators, always an eyesore, as in:

G1 = (a:b:c,x:(a,b))
Sure, use the colon and make it descending. There, we satisfied the schema folks.

The other topic, data flow with SQL engines. Here is how I see it.

For the most part when considering @(G1,G2), the reality will be a huge data farm with enormous lists of ontology in G2 and a small user ontology for matching. The user had 50 words, the farm has 300,000 is more likely.

Well, like the colon operator above, the huge farm will create special operators, on its own machines. Theire operators functions obey simple G grammar and the machine handles the rest.

Remember, everything is a triple: [keyword,link,pointer] in G world. All the good information is in keyword, but tons of info in the arrangement of links, and pointer arithmetic makes for very optimum searches. And, by the way, it adds value to SQL programming expertise.

But, the data flow again. An SQL programmer may have this thin tied up for milliseconds, doing on operation, but the one operation is tied to a complex form like:

insert into result select from G1 and G2 on keyword, link function and maybe property, rank (descending order). Complex, maybe collection 1,000 words from one list as the intersection with the user list, all in G graph form. But, it stops, at some point the insert has run the course, and the current graph, the one who guides the process in nested graph order, that graph, yields the semaphore. But it leaves the stop condition on top of the result table. Then the G mahine takes over, it pops the recodr off of result and deletes. The next step is generally a link, or a condition that requires the opposing graph, which then gets the machine.
file (sibling property)

No comments: