def creatMultiItemUserAdj(dataset, cv):
trainMat, _, _, trainMat_time, _ = loadData(dataset, cv)
ratingClass = np.unique(trainMat.data).size
userNum, itemNum = trainMat.shape
multi_adj = sp.lil_matrix((ratingClass * itemNum, userNum), dtype=np.int)
uidList = trainMat.tocoo().row
iidList = trainMat.tocoo().col
rList = trainMat.tocoo().data
# time = trainMat_time.tocoo().data
for i in range(uidList.size):
uid = uidList[i]
iid = iidList[i]
r = rList[i]
multi_adj[iid * ratingClass + r - 1, uid] = trainMat_time[uid, iid]
assert trainMat_time[uid, iid] != 0
a = sp.csr_matrix((multi_adj.shape[1], multi_adj.shape[1]))
b = sp.csr_matrix((multi_adj.shape[0], multi_adj.shape[0]))
multi_adj2 = sp.vstack(
[sp.hstack([a, multi_adj.T]),
sp.hstack([multi_adj, b])])
DIR = os.path.join(os.getcwd(), "dataset", dataset, 'implicit',
"cv{0}".format(cv))
path = DIR + '/multi_item_adj.pkl'
with open(path, 'wb') as fs:
pickle.dump(multi_adj2.tocsr(), fs)
print("create multi_item_feat")
# if __name__ == '__main__':
# creatMultiItemUserAdj("CiaoDVD_time", 1)
# print("Done")
def getData(self, args):
trainMat = loadData(args.dataset, args.cv)
with open(self.datasetDir + '/uu_vv_graph.pkl', 'rb') as fs:
uu_vv_graph = pickle.load(fs)
uuMat = uu_vv_graph['UU'].astype(np.bool)
iiMat = uu_vv_graph['II'].astype(np.bool)
return trainMat, uuMat, iiMat
def getData(self, args):
data = loadData(args.dataset)
trainMat, _, validData, _, _ = data
with open(self.datasetDir + '/multi_item_adj.pkl', 'rb') as fs:
multi_adj_time = pickle.load(fs)
with open(self.datasetDir + '/uu_vv_graph.pkl', 'rb') as fs:
uu_vv_graph = pickle.load(fs)
uuMat = uu_vv_graph['UU'].astype(np.bool)
iiMat = uu_vv_graph['II'].astype(np.bool)
return trainMat, validData, multi_adj_time, uuMat, iiMat
def getData(self, args):
trainMat, testMat, validMat, trustMat = loadData(
args.dataset, args.rate)
a = trainMat + testMat + validMat
assert a.nnz == (trainMat.nnz + testMat.nnz + validMat.nnz)
adj_DIR = os.path.join(os.getcwd(), "data", dataset, 'mats')
adj_path = adj_DIR + '/{0}_multi_item_adj.pkl'.format(args.rate)
with open(adj_path, 'rb') as fs:
multi_adj = pickle.load(fs)
return trainMat, testMat, validMat, trustMat, multi_adj
def creatMultiItemUserAdj(dataset, rate):
trainMat, _, _, _ = loadData(dataset, rate)
ratingClass = np.unique(trainMat.data).size
userNum, itemNum = trainMat.shape
mult_adj = sp.lil_matrix((ratingClass * itemNum, userNum), dtype=np.int)
uidList = trainMat.tocoo().row
iidList = trainMat.tocoo().col
rList = trainMat.tocoo().data
for i in range(uidList.size):
uid = uidList[i]
iid = iidList[i]
r = rList[i]
mult_adj[iid * ratingClass + r - 1, uid] = 1
DIR = os.path.join(os.getcwd(), "data", dataset, 'mats')
path = DIR + '/{0}_multi_item_adj.pkl'.format(rate)
with open(path, 'wb') as fs:
pickle.dump(mult_adj.tocsr(), fs)
print("create multi_item_feat")
def getData(self, args):
data = loadData(args.dataset)
trainMat, testData, validData, trustMat = data
return trainMat, testData, validData, trustMat