Friday, October 28, 2011

Ontology reduction rules

I suggested one, cropping euivalent sets of key word to have uniform clikc thrus.

What do we vall i when we preced a graph with a keyword?
Sep 28, 1997 becomes data.Sep 28, 1997.(URL1,URL2,..)
where we have indexed text according to the appearance of the data. Dates can be kept as literals or in computer dates/time format. The user types anything looking like a date format, the search returns url pointers, ordered by click thru regardless of literal for formatted match.

So, preceding a set of keywords with a prefix is indexing: date.date/time literal.(url pointers)
Using my simle syntax. So we can develop ten or so rules for graph morphing, always yield string sets such the they are maximum entropy, measured by click thrus. We can describe the graph morphing with some standard algebra. (All this is likely done somewhere in whole or part).

No comments: