More excellent mind-blowing thinking today.
Google is taking the technology that it used to generate Google Translate (which doesn’t really understand languages) and applying it to thoughts.
“If you take the vector for Paris and subtract the vector for France and add Italy, you get Rome,†he said. “It’s quite remarkable.â€
Source: Google Is Working On A New Algorithm – Thought Vectors
This kind of vector analysis is fairly well used in fancy and expensive search systems such as Autonomy’s. I have heard it described as “document space” and “concept mapping” before. I know a bit about it because we once commissioned some original research on applying the technique to our concept space. The idea was to be able to mathematically contextualise term sequences, parts of documents and whole documents in terms of an n-dimensional space where each dimension is some kind of concept. You can then determine the “nearness” of one document to another or indeed an input such as a natural language query or just the expression of a thought. “Thought vectors” is a much sexier term! With Google’s corpus and resources I expect they will do a much better job. I guess contextualisation is one job that the human brain is superbly organised for, with its capacity for associative processing.
If you take the vector for 100 U.S.$ and subtract the same in debt does this equal cash?
Frazer hopefully, but only the thought of cash.