gistfile1.rb

I would start by running the test stuff.

test = euclidean(@people, 'Lisa Rose', 'Gene Seymour')
puts "#{test} should equal 0.148148148148"

 Also note that the version in the book is wrong,
which is why i have the #errata comment there, since my version is the correct one, afaik.

so, both euclidean and pearson are meant to be interchangeable. 
So, all that both of them do is take a hash in the following form:


{
  :owner_object => {:owned_object => score, :owned_object => score,
   :another_owner_object => {:owned_object => score}
}

Where score is an int. To try this for yourself, just construct some hash and 
pass in two of the keys.

So, euclidean will just take a hash like that, plus two of the keys 
for the owner objects and tell you how similar they are.

Does that make sense?

I wrote the method to be very clear for people / movies, 
but it's unfortunaly kind of unclear if you want to abstract it.
 However, it is abstract.


You can see all this at work in the delicious_recommendations.rb file. 

So, when it calls super, in top matches in the class in that file, it's calling
the top_matches defined in recommendations.rb, (since it's defined on object it gets called
by super -- i was lazy).

So, looking here, we can see that peearson is getting called.

I think that you could just change this to euclidean and it would still work. So that's another example.

def top_matches(people, person, n=5)
  scores = people.map do |critic, items|
    unless person == critic
      [(block_given? ? yield(people, person, critic) : pearson(people, person, critic)), critic]
    end
  end
 
  scores.compact.sort.reverse[0..n]
 
end

Best,

M