About GEO-PAGES

The idea for GEO-PAGES was spawned out of a simple thinking: "I want to see what's popular in my neighborhood." To create a site to accomplish this, - without years of data entry - means we must pull information about what's popular, and where it's located. Naturally, Facebook is the wellspring.

GEO-PAGES pulls publically accessible information about pages created, and managed, on Facebook. Since it wouldn't be necessary (or feasible) to pull EVERY Facebook page at once, pages are fetched from facebook as they are discovered through users of GEO-PAGES. Basically, this means that when you grant GEO-PAGES access to your facebook information, the site is able to learn about the things that you like. In turn, this learned information is used to offer suggestions to GEO-PAGES users, and to determine the similarity of pages within the system.

GEO-PAGES is a small project. It may function slowly at times, and automatic suggestions may take some time to generate. Please be considerate. If you interested in the project, hit me up. I NEED to hear YOUR feedback.

Absolutely:

  • GEO-PAGES will never give your information to anyone. Ever.
  • GEO-PAGES will never contact you. Ever.
  • GEO-PAGES will never cost money. Ever.

How suggestions work:

For the technically curious, GEO-PAGES currently uses two methods (in combination) to generate suggested pages for each user.

  1. Page Similarity
    • The similarity of pages is determined. This is accomplished by recording a 'handshake' between two pages. Whenever a user likes both pages, the bond between those pages is strengthened. As data is added to the system, this method becomes an accurate means of measuring the similarity between any two pages.
    • The highest ranked pages, based on your likes are suggested. Iterating through your current likes, a number of the most similar pages are compiled. These pages are ordered by the strength of their bond to your current likes (and therefore to you), and indexed as suggestions.
    • This method generates suggestions quickly, but tends to suggest popular things, overlooking 'neighbourhood' pages
  2. Friend's Suggestions
    • Your most similar friends are determined. Iterating through your friends, a bond-strengh is determined based on the number of shared likes.
    • Your closest friends have more voting power. Your most similar friends are awarded an appropriate degree of voting power, to appropriately weight suggestions in the following step.
    • The most 'popular' pages are suggested. Iterating through each of your top friends' likes, the highest voted pages are indexed as suggestions.
    • This method takes longer to generate, but performs well at suggesting 'local', less mainstream pages that you are likely to appreciate.