Wednesday, March 1, 2017

I invent recipe bot

Recipe bot is a semantic graph convolution process.  All the foods are connect by cooking groups, a semantic garph that partitions according to similarly derived food groups.

The search graph is a set of cooking rules with two or more foods. The user enters the food groups in the desired recipe, the recipe is constructed on the fly as an output graph from the convolution of the search and semantic graphs.

For you monotonarians, a recipe is like an unpriced balance sheet. But it is multi-color, volume and cook time vary somewhat independently, so recipes end up as triple entry account.

I call this search method, in general, the unpatentable process of semantic graph matching.

Consider your basic cookie

Unleavened bread based in wheat flour is a distinct sub set, and its is a discernible, meaning discoverable, subset that is graphed vie general reconsructio of a whole set of recipes.  Or just follow the recipe web site from one of the unleavened flour sellerw, the web sit is organixed by constituent recipe components.

But, the base mi for cookie is relatively the same, it branches out according to included spices and treats, as well as a variance in cooking time and relative volumes.  But thwe typical ixes at each branch point are known, they form the branch point.

Recipe rules are a branch ans set between different mixing/chopping and cooking sequences.

From an automated learning problem, the main issue is find the frequency of use of the various known reipes.  Otherwise, as noted above, recipes already are organized for reconstruction.

Price, by the way, is already a strong component of the online recipes. On the web they compe constructed with priced products, and often are thus organized for optimum pricing.  So we can see the overlap between priced and unpriced semantic graphs.

Relative to the sandbox

Recipe bot is above the pure cash layer, it is moving mass over time and distance.  From the perspective of WalMart, recipe bot applications are great for family shoppers, because Walmart inventory can track the most popular bot recipes; the app makes statistics available to pure cash. So we get the virtuous circle, the recipe bot app works at the mass layer via coordination, and snookers the S&Ls into a requant in favor of recipe users.  That is called productive growth.

No comments: