Partial ordering again

How do we use it in lazy J?
given: G = "k1 k2 k3 k4 k5"  quoted, so the only operator is a weak semi-ordering.

This graph {k1,k2,k3}.{k2,k3,k4},{k3,k4,k5}  // Windowed
K3 evidently is a very common word in the general semantic chain. But this is very likely to correspond to near grammar straight from a key word filter.  There should exist a key word expander for G, @G,X
that generates possible expression based on prior frequency of use measurements. So we can imagine X, in the expression is a mixture of wild cards and perhaps some lightweight Json arithmetic.  So we can add likelihood coefficients determine our And match,pass and collect selection.  Wow!, How about that! Some lightweight Json arithmetic.  I have some prior work in this area, stay tuned.

Ok, cnside a micro map on k, Vk, gives an accummulant I can add as in: a += Vk for any given key word k, at this location in the graph.  Remember, location in the grammar is always implicit.  how's that look?

A > V{k1,k2,k3}.V{k2,k3,k4}  // Keep going until found or accumulant exhausted.

The micro map V distributes, and each k its own value. Vk1 match Vk2 the closeness being a value from a few variants.  "hair fur feather clothing" feather and clothing might be a weak match.    Assigning variable space and pointers in nested store? A breeze for a light weight arithmetic, an intelligent store manager, and something variable like a triplet. Make space available in the link value for overloaders to add small bit count weightings. Make a better Watson.