I wasn't working on it for 3 months and so there was no need because my pool of sites was quite small and my approach was more oriented for testing a number of variables. Times have changed as it usually is every few months and distraction happens to be life's specialty. Now, what I need is what I spoke about back then. Create websites which all have unique content and look like created by different entities. The first problem is that a crawler can read my website visually. So that's the first and major problem. Another problem is that my websites that look similar can be connected and can be assumed to be created by the same person if they're on free subdomains.
Now one solution is to create a tricky cloudflare captcha and it would mimic real captcha. However, this doesn't guarantee unique DOM on all of my sites which leaves area for interconnection and suspicion from data analyzed by algorithms gathered by crawlers. To undertake a task of creating a site creator, there needs to be a
64 bit compiler that does as follows:
- Leverages database of images, styles and unique text to substitute in templates.
- A database has 7 tables: image src, text, keywords, person, page, page_style, element_types.
- A single person has many pages which consist of references to keywords, images and texts.
- A single person has one page_style which consists of columns: image_proportions, image_dimensions, borders_enabled, shadows_enabled borders_weight, shadows_weight, dominant_shades, secondary_shades, tertiary_shades, background_color, colors_or_gradients, amount_of_sentences_per_page, image_file_formats, dom_complexity, dimensions_units_type, colors_units_type, human_readable_classnames, utility_style_classes, font_get_style, scripts_placement, css_styles_placement.
- Element types have columns as follows: type, dom_complexity, width, height,
Steps to fill for each person in the database include:
- Finding images related to a keyword specified with dimensions proportions similar to ones in settings. Word "scraping" was changed on purpose.
- Finding text related to a keyword specified. It happens to be enough to have a number of sentences from a range of 2 - 20.
- Creating styles based on predefined settings which are chosen in a chaotic fashion, as a typical web designer would do.
- Leverages an algorithm to cut off excess of images to fit dimensions specified in page_style table.
- Leverages an algorithm to analyze data in database to create a template according to details specified for a certain person.
- Leverages an algorithm to substitute all data for an amount of keywords in a template according to behavior specified in functions procedures.
In summary:
- Puts DOM elements in place with random classes and assigns CSS.
- Creates a visually unique template.
- Substitutes variables in templates to create a set of 50+ pages under 165 milliseconds.
- Helps pretend to be a normal site that has nothing to do with any marketing activity, but is made to serve only white hat purposes.
Now that's a draft subject to changes.
