I'd like to see a more intuitive visual, such as:
Say the algorithm is a compression algorithm, that builds a tree of seen sentences (of 0-14 letters in length, all them it came across). And updates its nodes how many times seen each ex. 44 times for 'we walked', 2749 times for 'walk', etc. It then goes over a file and compresses it by predicting the likely next letter but for its error it stores a little correction to its comp pad.
So, the visual would be a big page of writing - the text file........and a little bar sliding across it sending off branch updates to a tree that counts times seen each piece......Then, when done, the other text file has a bar go across it and the bar snippet snapshotted is sent into the tree and it drops off predictions like a falling leaf to the ground, and these leaves then meet with a number thing to get the correction number, of which is added to a long pad being the compression text pad page, another 3rd page of text.
How to do it? Well you have the 3 files as input, the text to train on, the text to compress, and the output pad page. And you are selecting parts of the file, which can be visualized I think. Where these go is into a tree in this example, but how do we get a tree from a list system made up in the human's brain is a "tree" but not visually a tree? It's just a list, but it operates as a tree!! Maybe we can make something to make a visual from it! So say the first item in the list is always came across, all 256 of them, a, b, c, and so on, and so these are put into a row...the first layer. Then, the template algorithm sees ok so every time the item a is seen it sees then items in the list only at these next 256 possible positions, no where else at all! So, it associates them....this would look like a tree, each node connecting to its children accesses, the first letter of a text snippet does not access the (at least in this example) the 3rd letter ex. 'wa[l]k', so it does not jump to that item in the long list.
But what about the IF THEN? And AND? Etc? I think if we can show the main files, and structures that aren't there but are in the human's brain, we can already visualize LOTS of what's going on.
Just an idea, but, it may work!
