i'm working on a project that tries to match the events in a city based on the interests of the user. Basically, the end user in the front end has to select his interests (culture, sports, fun, food...there are 8 of them) with values that range from 0 to 10. In the back end (where all the events are created) the webmaster can choose a score for each category (culture, sports, fun...the same as the front end). What is the best algorithm to use to find the most relevant results based on the user input? I'm using php (although i think this is not a language specific question).
Compare each of the scores of the events with those from the user profile. The lower the deviation, the better the event matches the user's interest.
In SQL it could be something like this:
SELECT ... FROM `events` WHERE ...
ORDER BY (
ABS(`culture` - $culture)
+ ABS(`sports` - $sports)
+ ABS(`fun` - $fun)
+ ...
) ASC
You could also weight the scores according to the distribution of points in the user profile. That means if a user has $culture = 10 and all other scores 0, he is particularly interested in culture and you may get better results with
SELECT ... FROM `events` WHERE ...
ORDER BY (
(ABS(`culture` - $culture) * $cultureImportance)
+ (ABS(`sports` - $sports) * $sportsImportance)
+ (ABS(`fun` - $fun) * $funImportance)
+ ...
) ASC
where the importance of each score is the deviation of that score from the mean of all scores of the user.
You could also add "spice" to the importance values, e.g. to push sport events during the time of superbowl to make your site more zeitgeisty.
I don't know about an algoritm, but you can multiply the users interest values with the events scores and and add them all per event. Thiss way you can rank the events for the user.
For example User 1 A : 8 B : 6 C : 0
User 2 A : 2 B : 8 C : 6
Event 1 A : 8 B : 0 C : 5
Event 2 A : 2 B : 6 C : 0
User 1 event hitparade Event 1 : 64 Event 2 : 52
User 2 event hitparade Event 2 : 52 Event 1 : 46