We're looking for a mathematician

Hello, We have texts composed of sections. In these sections, we sometimes have titles but we always have paragraphs. Inside paragraphs, we have one or several sentences. Sometimes, in one section, we have what we name "blocks". A block is a little group made of one title and one or several paragraphs. We are developing a tool and we need a mathematician to help us take the best decision. With this tool, users will import a batch of several hundreds or thousands of texts, gathered in 1 file. All these texts will have the same structure (same number of sections, blocks, titles, paragraphs and sentences, all in the same order. Users will then define: 1/ Select only a part of the elements: ● If we must use all sections for the output texts or if the tool must use between x and y sections upon the total number of sections that we have in the origin texts. ● Same with blocks inside sections. ● Same with paragraphs inside blocks and sections. ● Same with sentences inside paragraphs. ● If we can sometimes hide the title of a block or if we must "print" it in each and every output text. 2/ Swap some elements: ● If we can swap the sections to get them in a different order in each and every output text. ● Same with blocks inside sections. ● Same with paragraphs inside blocks and sections. ● Same with sentences inside paragraphs. If we have 37 456 texts in the input file, we must get 37456 texts in the output file. What we want is to get the most different structures in output between each text. We think this can be achieved by considering that each sentence/paragraph/title/block/section is a distinct element in a sequence. The goal would then be to use the principle of Hamming distance, to get the most different sequences in the output texts. But if you think there's a better way to achieve this goal, we're all ears. 1st question you will have to answer to: is it better to work with smaller sequences (one sequence for one paragraph, then one sequence for one block, then one sequence for one section, then one sequence to select and swap sections) or is it better to work sequences globally, gathering all the elements for 1 text inside 1 longer sequence? 2nd question: are you able to code the algorithm? (it's not mandatory, as long as you can explain the principles to a developer). Best regards, Marco.