# 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
import scipy.sparse as sp
from cornac.models.c2pf import c2pf
from ..recommender import Recommender
from ..recommender import ANNMixin, MEASURE_DOT
# Recommender class for Collaborative Context Poisson Factorization (C2PF)
[docs]
class C2PF(Recommender, ANNMixin):
"""Collaborative Context Poisson Factorization.
Parameters
----------
k: int, optional, default: 100
The dimension of the latent factors.
max_iter: int, optional, default: 100
Maximum number of iterations for variational C2PF.
variant: string, optional, default: 'c2pf'
C2pf's variant: c2pf: 'c2pf', 'tc2pf' (tied-c2pf) or 'rc2pf' (reduced-c2pf). \
Please refer to the original paper for details.
name: string, optional, default: None
The name of the recommender model. If None, \
then "variant" is used as the default name of the model.
trainable: boolean, optional, default: True
When False, the model is not trained and Cornac assumes that the model already \
pre-trained (Theta, Beta and Xi are not None).
Item_context: See "cornac/examples/c2pf_example.py" in the GitHub repo for an example of how to use \
cornac's graph modality to load and provide "item context" for C2PF.
init_params: dict, optional, default: None
List of initial parameters, e.g., init_params = {'G_s':G_s, 'G_r':G_r, 'L_s':L_s, 'L_r':L_r, 'L2_s':L2_s, 'L2_r':L2_r, 'L3_s':L3_s, 'L3_r': L3_r}
Theta: ndarray, shape (n_users, k)
The expected user latent factors.
Beta: ndarray, shape (n_items, k)
The expected item latent factors.
Xi: ndarray, shape (n_items, k)
The expected context item latent factors multiplied by context effects Kappa.
G_s: ndarray, shape (n_users, k)
Represent the "shape" parameters of Gamma distribution over Theta.
G_r: ndarray, shape (n_users, k)
Represent the "rate" parameters of Gamma distribution over Theta.
L_s: ndarray, shape (n_items, k)
Represent the "shape" parameters of Gamma distribution over Beta.
L_r: ndarray, shape (n_items, k)
Represent the "rate" parameters of Gamma distribution over Beta.
L2_s: ndarray, shape (n_items, k)
Represent the "shape" parameters of Gamma distribution over Xi.
L2_r: ndarray, shape (n_items, k)
Represent the "rate" parameters of Gamma distribution over Xi.
L3_s: ndarray
Represent the "shape" parameters of Gamma distribution over Kappa.
L3_r: ndarray
Represent the "rate" parameters of Gamma distribution over Kappa.
References
----------
* Salah, Aghiles, and Hady W. Lauw. A Bayesian Latent Variable Model of User Preferences with Item Context. \
In IJCAI, pp. 2667-2674. 2018.
"""
def __init__(
self,
k=100,
max_iter=100,
variant="c2pf",
name=None,
trainable=True,
verbose=False,
init_params=None,
):
if name is None:
Recommender.__init__(
self, name=variant.upper(), trainable=trainable, verbose=verbose
)
else:
Recommender.__init__(self, name=name, trainable=trainable, verbose=verbose)
self.k = k
self.max_iter = max_iter
self.ll = np.full(max_iter, 0)
self.eps = 0.000000001
# self.aux_info = aux_info # item-context matrix in the triplet sparse format: (row_id, col_id, value)
self.variant = variant
# Init params if provided
self.init_params = {} if init_params is None else init_params
self.Theta = self.init_params.get("Theta", None)
self.Beta = self.init_params.get("Beta", None)
self.Xi = self.init_params.get("Xi", None)
self.Gs = self.init_params.get("G_s", None)
self.Gr = self.init_params.get("G_r", None)
self.Ls = self.init_params.get("L_s", None)
self.Lr = self.init_params.get("L_r", None)
self.L2s = self.init_params.get("L2_s", None)
self.L2r = self.init_params.get("L2_r", None)
self.L3s = self.init_params.get("L3_s", None)
self.L3r = self.init_params.get("L3_r", None)
[docs]
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)
X = train_set.csr_matrix
# recover the striplet sparse format from csc sparse matrix X (needed to feed c++)
(rid, cid, val) = sp.find(X)
val = np.array(val, dtype="float32")
rid = np.array(rid, dtype="int32")
cid = np.array(cid, dtype="int32")
tX = np.concatenate(
(np.concatenate(([rid], [cid]), axis=0).T, val.reshape((len(val), 1))),
axis=1,
)
del rid, cid, val
if self.trainable:
# use pre-trained params if exists, otherwise from constructor
init_params = {
"G_s": self.Gs,
"G_r": self.Gr,
"L_s": self.Ls,
"L_r": self.Lr,
"L2_s": self.L2s,
"L2_r": self.L2r,
"L3_s": self.L3s,
"L3_r": self.L3r,
}
train_item_indices = set(train_set.uir_tuple[1])
(rid, cid, val) = train_set.item_graph.get_train_triplet(
train_item_indices, train_item_indices
)
context_info = np.hstack(
(rid.reshape(-1, 1), cid.reshape(-1, 1), val.reshape(-1, 1))
)
if self.variant == "c2pf":
res = c2pf.c2pf(
tX,
X.shape[0],
X.shape[1],
context_info,
X.shape[1],
X.shape[1],
self.k,
self.max_iter,
init_params,
)
elif self.variant == "tc2pf":
res = c2pf.t_c2pf(
tX,
X.shape[0],
X.shape[1],
context_info,
X.shape[1],
X.shape[1],
self.k,
self.max_iter,
init_params,
)
elif self.variant == "rc2pf":
res = c2pf.r_c2pf(
tX,
X.shape[0],
X.shape[1],
context_info,
X.shape[1],
X.shape[1],
self.k,
self.max_iter,
init_params,
)
else:
res = c2pf.c2pf(
tX,
X.shape[0],
X.shape[1],
context_info,
X.shape[1],
X.shape[1],
self.k,
self.max_iter,
init_params,
)
self.Theta = sp.csc_matrix(res["Z"]).todense()
self.Beta = sp.csc_matrix(res["W"]).todense()
self.Xi = sp.csc_matrix(res["Q"]).todense()
# overwrite init_params for future fine-tuning
self.Gs = np.asarray(res["G_s"])
self.Gr = np.asarray(res["G_r"])
self.Ls = np.asarray(res["L_s"])
self.Lr = np.asarray(res["L_r"])
self.L2s = np.asarray(res["L2_s"])
self.L2r = np.asarray(res["L2_r"])
self.L3s = np.asarray(res["L3_s"])
self.L3r = np.asarray(res["L3_r"])
elif self.verbose:
print("%s is trained already (trainable = False)" % (self.name))
return self
[docs]
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 which 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 self.variant == "c2pf" or self.variant == "tc2pf":
if item_idx is None:
user_pred = (
self.Beta * self.Theta[user_idx, :].T
+ self.Xi * self.Theta[user_idx, :].T
)
else:
user_pred = (
self.Beta[item_idx, :] * self.Theta[user_idx, :].T
+ self.Xi * self.Theta[user_idx, :].T
)
elif self.variant == "rc2pf":
if item_idx is None:
user_pred = self.Xi * self.Theta[user_idx, :].T
else:
user_pred = self.Xi[item_idx,] * self.Theta[user_idx, :].T
else:
if item_idx is None:
user_pred = (
self.Beta * self.Theta[user_idx, :].T
+ self.Xi * self.Theta[user_idx, :].T
)
else:
user_pred = (
self.Beta[item_idx, :] * self.Theta[user_idx, :].T
+ self.Xi * self.Theta[user_idx, :].T
)
# transform user_pred to a flatten array,
user_pred = np.array(user_pred, dtype="float64").flatten()
return user_pred
[docs]
def get_vector_measure(self):
"""Getting a valid choice of vector measurement in ANNMixin._measures.
Returns
-------
measure: MEASURE_DOT
Dot product aka. inner product
"""
return MEASURE_DOT
[docs]
def get_user_vectors(self):
"""Getting a matrix of user vectors serving as query for ANN search.
Returns
-------
out: numpy.array
Matrix of user vectors for all users available in the model.
"""
if self.variant == "rc2pf":
user_vectors = np.concatenate((self.Theta, self.Theta), axis=1)
else:
user_vectors = self.Theta
return user_vectors
[docs]
def get_item_vectors(self):
"""Getting a matrix of item vectors used for building the index for ANN search.
Returns
-------
out: numpy.array
Matrix of item vectors for all items available in the model.
"""
if self.variant == "rc2pf":
item_vectors = np.concatenate((self.Beta, self.Xi), axis=1)
else:
item_vectors = self.Beta
return item_vectors
