A reader asked me how he could apply back mapping (BM) to SEO work.

Essentially, any algorithm like BM that works at the level of collections can be scaled down and used at a page level. One just needs to decompose a page into passages (chunks of text or text windows) and treat each passage as a pseudo document.

First, convert a document into a text stream; i.e., linearize, tokenize, and filtrate the document. You should end with a linearized piece of text.

 If it turns out that the linearized document is almost identical to the original document,  you might want to go back and partition the original document every x paragraphs or sentences. If not, partition the text stream so you end with segments or text windows. You need to decide if you want to do this very x number of words or characters.

Next, treat each segment or chunk of text as a “document” and construct a term-“document” matrix “A”.

If you want to narrow this down to the level of links, construct a term-link matrix.

 In general, an algorithm that applies to a true term-document matrix A can be applied to “A”.

Welcome to scaling.