Feed Watson all your favorite investment books, for a starter. Then give it PoA. Watson watches all the markets, all of them fair traded transparent auto-priced. Watson went to college and knows all the formal syntax and semantics of the cash layer, the contracts traded. Watson can handle risky bets.
So Watson has a huge semantic graph, and your investment profile is a much smaller search graph. The convolution of the two yields your investment recommended. The result also a semantic graph, in JSON, fed to your card. The card handles the rest.
That was the point of the semantic machine project. Consider convolving two JSON directed graphs. Add your twenty rules of grammar, and changed them in other context, but you just kept this core JSON convolving machine. All attachments had the same basic interface, it had to be presented as a graph responding to four or five basic graph traversals.
Javascript simply has Javascript grammar, but otherwise is a JSON convolution machine. Take the core, define your special grammar, then add e book, financial data, all in run time, just make it look like a directed graph. Like a component system, at is core is the generic JSON convolution. Then you get a run time pattern match system, allowing ad hoc rules, as needed for contex. A machine like this, a Watson, can watch the pure cash markets, for a fee. We are all going to get our own Warson, soon.
No comments:
Post a Comment