util/sampler.py

from random import shuffle,randint,choice

# training_data:[[user_id, item_id, float(weight)]...]






def next_batch_pairwise(data,batch_size,n_negs=1):
    """

    args:
        data
        batch_sieze
        n_negs = 64


    return:
        u_idx : [batch_size]
        i_idx : [batch_size]
        j_idx : [batch_size,64]

    """
    training_data = data.training_data
    # 传进来的参数data是data.ui_graph下的Interaction类
    shuffle(training_data)
    batch_id = 0
    data_size = len(training_data)
    while batch_id < data_size:
        if batch_id + batch_size <= data_size:
            users = [training_data[idx][0] for idx in range(batch_id, batch_size + batch_id)]
            items = [training_data[idx][1] for idx in range(batch_id, batch_size + batch_id)]
            batch_id += batch_size
        else:
            users = [training_data[idx][0] for idx in range(batch_id, data_size)]
            items = [training_data[idx][1] for idx in range(batch_id, data_size)]
            batch_id = data_size
        u_idx, i_idx, j_idx = [], [], []
        item_list = list(data.item.keys()) # 以列表的方式来获取所有item_id

        for i, user in enumerate(users):
            i_idx.append(data.item[items[i]])
            u_idx.append(data.user[user])
            neg_idx=[]
            for m in range(64):
                neg_item = choice(item_list)
                while neg_item in data.training_set_u[user]:
                    neg_item = choice(item_list)
                neg_idx.append(data.item[neg_item])
                # 这里的nef_item_id的shpe应该是[batch,64]
            j_idx.append(neg_idx)
        yield u_idx, i_idx, j_idx


def next_batch_pointwise(data,batch_size):
    training_data = data.training_data
    data_size = len(training_data)
    batch_id = 0
    while batch_id < data_size:
        if batch_id + batch_size <= data_size:
            users = [training_data[idx][0] for idx in range(batch_id, batch_size + batch_id)]
            items = [training_data[idx][1] for idx in range(batch_id, batch_size + batch_id)]
            batch_id += batch_size
        else:
            users = [training_data[idx][0] for idx in range(batch_id, data_size)]
            items = [training_data[idx][1] for idx in range(batch_id, data_size)]
            batch_id = data_size
        u_idx, i_idx, y = [], [], []
        for i, user in enumerate(users):
            i_idx.append(data.item[items[i]])
            u_idx.append(data.user[user])
            y.append(1)
            for instance in range(4):
                item_j = randint(0, data.item_num - 1)
                while data.id2item[item_j] in data.training_set_u[user]:
                    item_j = randint(0, data.item_num - 1)
                u_idx.append(data.user[user])
                i_idx.append(item_j)
                y.append(0)
        yield u_idx, i_idx, y