-
Notifications
You must be signed in to change notification settings - Fork 140
/
recom_hft.py
200 lines (154 loc) · 7.54 KB
/
recom_hft.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
# Copyright 2018 The Cornac Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
import numpy as np
from ..recommender import Recommender
from ...exception import ScoreException
class HFT(Recommender):
"""Hidden Factors and Hidden Topics
Parameters
----------
name: string, default: 'HFT'
The name of the recommender model.
k: int, optional, default: 10
The dimension of the latent factors.
max_iter: int, optional, default: 50
Maximum number of iterations for EM.
grad_iter: int, optional, default: 50
Maximum number of iterations for L-BFGS.
lambda_text: float, default: 0.1
Weight of corpus likelihood in objective function.
l2_reg: float, default: 0.001
Regularization for user item latent factors.
vocab_size: int, optional, default: 8000
Size of vocabulary for review text.
init_params: dictionary, optional, default: None
List of initial parameters, e.g., init_params = {'alpha': alpha, 'beta_u': beta_u,
'beta_i': beta_i, 'gamma_u': gamma_u, 'gamma_v': gamma_v}
alpha: float
Model offset, optional initialization via init_params.
beta_u: ndarray. shape (n_user, 1)
User biases, optional initialization via init_params.
beta_u: ndarray. shape (n_item, 1)
Item biases, optional initialization via init_params.
gamma_u: ndarray, shape (n_users,k)
The user latent factors, optional initialization via init_params.
gamma_v: ndarray, shape (n_items,k)
The item latent factors, optional initialization via init_params.
trainable: boolean, optional, default: True
When False, the model will not be re-trained, and input of pre-trained parameters are required.
verbose: boolean, optional, default: True
When True, some running logs are displayed.
seed: int, optional, default: None
Random seed for weight initialization.
References
----------
Julian McAuley, Jure Leskovec. "Hidden Factors and Hidden Topics: Understanding Rating Dimensions with Review Text"
RecSys '13 Proceedings of the 7th ACM conference on Recommender systems Pages 165-172
"""
def __init__(self, name='HFT', k=10, max_iter=50, grad_iter=50,
lambda_text=0.1, l2_reg=0.001, vocab_size=8000,
init_params=None, trainable=True, verbose=True, seed=None):
super().__init__(name=name, trainable=trainable, verbose=verbose)
self.k = k
self.lambda_text = lambda_text
self.l2_reg = l2_reg
self.grad_iter = grad_iter
self.name = name
self.max_iter = max_iter
self.verbose = verbose
self.init_params = {} if not init_params else init_params
self.seed = seed
self.vocab_size = vocab_size
def fit(self, train_set, val_set=None):
"""Fit the model to observations.
Parameters
----------
train_set: :obj:`cornac.data.Dataset`, required
User-Item preference data as well as additional modalities.
val_set: :obj:`cornac.data.Dataset`, optional, default: None
User-Item preference data for model selection purposes (e.g., early stopping).
Returns
-------
self : object
"""
Recommender.fit(self, train_set, val_set)
from ...utils.init_utils import normal
self.n_item = self.train_set.num_items
self.n_user = self.train_set.num_users
self.alpha = self.init_params.get('alpha', train_set.global_mean)
self.beta_u = self.init_params.get('beta_u', normal(self.n_user, std=0.01, random_state=self.seed))
self.beta_i = self.init_params.get('beta_i', normal(self.n_item, std=0.01, random_state=self.seed))
self.gamma_u = self.init_params.get('gamma_u', normal((self.n_user, self.k), std=0.01, random_state=self.seed))
self.gamma_i = self.init_params.get('gamma_i', normal((self.n_item, self.k), std=0.01, random_state=self.seed))
if self.trainable:
self._fit_hft()
return self
@staticmethod
def _build_data(csr_mat):
index_list = []
rating_list = []
for i in range(csr_mat.shape[0]):
j, k = csr_mat.indptr[i], csr_mat.indptr[i + 1]
index_list.append(csr_mat.indices[j:k])
rating_list.append(csr_mat.data[j:k])
return index_list, rating_list
def _fit_hft(self):
from .hft import Model
from tqdm import trange
# document data
bow_mat = self.train_set.item_text.batch_bow(np.arange(self.n_item), keep_sparse=True)
documents, _ = self._build_data(bow_mat) # bag of word feature
# Rating data
user_data = self._build_data(self.train_set.matrix)
item_data = self._build_data(self.train_set.matrix.T.tocsr())
model = Model(n_user=self.n_user, n_item=self.n_item, alpha=self.alpha, beta_u=self.beta_u, beta_i=self.beta_i,
gamma_u=self.gamma_u, gamma_i=self.gamma_i, n_vocab=self.vocab_size, k=self.k,
lambda_text=self.lambda_text, l2_reg=self.l2_reg, grad_iter=self.grad_iter)
model.init_count(docs=documents)
# training
loop = trange(self.max_iter, disable=not self.verbose)
for _ in loop:
model.assign_word_topics(docs=documents)
loss = model.update_params(rating_data=(user_data, item_data))
loop.set_postfix(loss=loss)
self.alpha, self.beta_u, self.beta_i, self.gamma_u, self.gamma_i = model.get_parameter()
if self.verbose:
print('Learning completed!')
def score(self, user_idx, item_idx=None):
"""Predict the scores/ratings of a user for an item.
Parameters
----------
user_idx: int, required
The index of the user for whom to perform score prediction.
item_idx: int, optional, default: None
The index of the item for that to perform score prediction.
If None, scores for all known items will be returned.
Returns
-------
res : A scalar or a Numpy array
Relative scores that the user gives to the item or to all known items
"""
if item_idx is None:
if self.train_set.is_unk_user(user_idx):
raise ScoreException("Can't make score prediction for (user_id=%d)" % user_idx)
known_item_scores = self.alpha + self.beta_u[user_idx] + self.beta_i + self.gamma_i.dot(
self.gamma_u[user_idx, :])
return known_item_scores
else:
if self.train_set.is_unk_user(user_idx) or self.train_set.is_unk_item(item_idx):
raise ScoreException("Can't make score prediction for (user_id=%d, item_id=%d)" % (user_idx, item_idx))
user_pred = self.alpha + self.beta_u[user_idx] + self.beta_i[item_idx] + self.gamma_i[item_idx, :].dot(
self.gamma_u[user_idx, :])
return user_pred