Question

I have a Ruby on Rails model that has quite a few HABTM relationships. Essentially, i’m making a video sharing site themed around skateboarding for a client. Each video has many attirbutes through HABTM: Location, Skateboarders, tags, location, music, equipment, spot, region, etc… And it also has a few class attributes that are significant: Title, description, etc…

The idea here is my client would like a “Related Videos” feed by each video player, using all of the formerly mentioned descriptive data. Furthermore, the information should be weighted (similar titles take precedence over similar tags).

I was trying to find a good way implement Google SiteSearch to handle the grunt work, but can’t find a good search term syntax ie:

inurl: example.com/videos related:example.com/videos/4638872 

Which doesn’t actually work, unfortunately…

Answer 1

Solved this by using ferret to do a keyword explode search.

def related
    string = self.name + " " + self.tag_name + " " + self.skateboarder_name + " " + self.category_name + " " + self.editor_name
    #probably could have used "join" there...
    s = "*" + string.gsub(" ", "* OR *") + "*"
    relates = Video.find_with_ferret(s)
    #takes out duplicates and gices 10 to choose from
    while relates.count <= 10
        relates << Video.random
        relates = relates.uniq
    end
    #take out original video
    relates = relates - [Video.find(self.id)]
    #send back with "middle data"... for some reason i thought this might be more accurate....
    return relates[1..8]
end