friends recommender

friends_recommender.py

class FriendsRecommender(MRJob):

    def steps(self):
        return [self.mr(self.map_input, self.count_number_of_friends),
                self.mr(self.count_max_of_mutual_friends,
                    self.top_recommendations)]

    def map_input(self, key, line):
        '''
        Compute a cartesian product using nested loops
        for each friend in connection_list

        Input (source -> {“friend1”, “friend2”, “friend3”}):
            marcel,jonas,maria,jose,amanda

        Output {[source, friend1], -1};
               {[friend1, friend2], 1};):

            ["jonas", "marcel"] -1
            ["jonas", "maria"]  1
            ["jonas", "jose"]   1
            ["amanda", "jonas"] 1
            ["marcel", "maria"] -1
            ["jose", "maria"]   1
            ["amanda", "maria"] 1
            ["jose", "marcel"]  -1
            ["amanda", "jose"]  1
            ["amanda", "marcel"]    -1

        '''
        input = line.split(';')
        user_id, item_ids = input[0], input[1:]
        for i in range(len(item_ids)):
            f1 = item_ids[i]
            if user_id < f1:
                yield (user_id, f1), -1
            else:
                yield (f1, user_id), -1

            for j in range(i + 1, len(item_ids)):
                f2 = item_ids[j]
                if f1 < f2:
                    yield (f1, f2), 1
                else:
                    yield (f2, f1), 1