I should address how the two go together. let's drop the triple terminology for a moment. The machine unpacks and packs nested order very fast and will be increasingly embedded into sqlite3. So it provides primitive containers, searchable by key word and structure. The machine knows about bytes and 32 bit words; and its match function is robust and programmable. It can use simple schema and it has some perfectly ugly punctuation marks. Otherwise, the machine is a very efficient packer and unpacker, it is like the metal container box in which global trade moves.
The knowledge systems academics are trying to develop is something to be imposed upon the container system. The machine can represent that a Thanksgiving dinner contains turkey and apple pie, at the primitive level. But it cannot order a turkey dinner from the supermarket. The knowledge system have to deploy lightweight tools on top of the container. Thanksgiving dinner and its components might be shopping idea, and let the machine unpack some local stores.
What the machine can do is discover that thanksgiving searches are often followed by supermarket orders. On a basic pattern match level, it can discover containers on its own. When it does so, it moves web pages to make them unpack more efficiently, but it will not delivery a turkey, yet.
No comments:
Post a Comment