About the meta learner

SML

This is an implementation for our SIGIR 2020 paper: How to Retrain Recommender System? A Sequential Meta-Learning Method.

Contributors: Yang Zhang, Chenxu Wang, Fuli Feng, Xiangnan He

pytorch >= 1.2

numpy

Save as array,

train: period_num.npy (user_id,item_id)
test: period_num.npy (user_id,pos_item_id, neg_item_id)
Dataset URL: https://rec.ustc.edu.cn/share/1e40a9b0-c80a-11ea-b178-77f793cf5b55 (This is Yelp, Adressa will be uploaded later)

nohup python main_yelp.py --MF_epochs=1 --TR_epochs=1 --multi_num=10 > yelp_log.out &

or

python main_yelp.py --MF_epochs=1 --TR_epochs=1 --multi_num=10

nohup python main_news.py --MF_epochs=2 --TR_epochs=2 --multi_num=7 > yelp_log.out &

for Adressa, if we set the MF_epochs and TR_epochs same to paper (=1) , we can also get a similar performance if we adjust multi_num.

	class trainDataset_withPreSample(Dataset):
	'''
	this is Dataset type for train transfer, the input dataset has sampled enough
	neg item. the each epoch, will select on cloumn neg_item as neg item
	'''
	def __init__(self,input_dataset):
	super(trainDataset_withPreSample, self).__init__()
	self.all_data = copy.deepcopy(input_dataset)
	self.have_read = 0
	self.neg_flag = np.arange(1, self.all_data.shape[1])
	np.random.shuffle(self.neg_flag)
	self.neg_all = input_dataset.shape[1]-2
	self.used_neg_count = 0
	self.data_len = input_dataset.shape[0]

	def __len__(self):
	return self.all_data.shape[0]

	def __getitem__(self, idx):
	user = self.all_data[idx,0]
	item = self.all_data[idx,1]
	neg_item = self.all_data[idx,self.neg_flag[self.used_neg_count]]
	self.have_read += 1
	if self.have_read >= self.data_len:
	self.have_read = 0
	self.used_neg_count += 1
	if self.used_neg_count >= self.neg_all:
	np.random.shuffle(self.neg_flag)
	self.used_neg_count = 0
	return user,item,neg_item