We consider two stories related if the words in each of the stories are related. We sort words in a headline from least common to most common, and determine how often the less common words from each headline appear together. If we think that the least common words in the headline for story A closely match the least common words in the headline for story B then we consider it a match.
Your +/- feedback helps us to determine when words from separate headlines should be considered "related".
If you're curious about the general technique, apart from some implementation details this essentially uses the "bag of words" training model described by Peter Norvig in this video.
You can use a filter! Filters allow you to select only stories that match certain criteria, such as a keyword or category.
To apply filters, just fill out the filter/sort form under "Filters and Sorts" on the home page and click apply. To save a filter as a default, just check the box on the filter/sort form (requires a user account).
You can use a sort! Sorts allow you to change the order in which stories appear, without changing what you're viewing.
To apply sorts, just select your desired option from the dropdown on the home page (under "Filters and Sorts") and click apply. To save a sort as a default, just check the box on the filter/sort form (requires a user account).
