Constrained in Item Similarity

[otakbeku]

Hi, I read your paper and pretty interested in your way. I would like to know more about item correlation, do you have any assumption about why you limiting the number of items in item similarity? For example, if I had more than >2000 items, the rest of the item will not record in the dictionary, right?

   def _generate_item_similarity(self,train=None, save_path='./'):
        """
        calculate co-rated users between items
        """
        print("getting item similarity...")
        train = train or self.train_data_dict
        C = dict()
        N = dict()
        for idx, (u, items) in enumerate(train.items()):
            if idx%2000 == 12:
                print("proceeded: ", idx)
                break