BanditLib/Articles_generator_r2bandit.py at master · nirandiw/BanditLib · GitHub

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
import cPickle
import numpy as np
from util_functions import featureUniform, gaussianFeature, fileOverWriteWarning
from random import sample, randint, uniform
import json
import time
from random import *

class Article():
	def __init__(self, aid, atype, FV=None):
		self.id = aid
		self.type = atype
		self.featureVector = FV


class ArticleManager():
	def __init__(self, dimension, n_articles, ArticleGroups, FeatureFunc, argv, userFeature_theta, userFeature_beta ):
		self.signature = "Article manager for simulation study"
		self.dimension = dimension
		self.n_articles = n_articles
		self.ArticleGroups = ArticleGroups
		self.FeatureFunc = FeatureFunc
		self.thetaFunc = FeatureFunc
		self.betaFunc = FeatureFunc
		self.argv = argv

		self.signature = "A-"+str(self.n_articles)+"+AG"+ str(self.ArticleGroups)+"+TF-"+self.FeatureFunc.__name__

		self.userFeature_theta = userFeature_theta
		self.userFeature_beta = userFeature_beta
	def saveArticles(self, Articles, filename, force = False):
		with open(filename, 'w') as f:
			for i in range(len(Articles)):
				f.write(json.dumps((Articles[i].id, Articles[i].type, Articles[i].featureVector)) + '\n')


	def loadArticles(self, filename):
		articles = []
		with open(filename, 'r') as f:
			for line in f:
				aid, atype, featureVector = json.loads(line)
				articles.append(Article(aid, atype,  np.array(featureVector)))
		return articles

	#automatically generate masks for articles, but it may generate same masks
	def generateMasks(self):
		mask = {}
		for i in range(self.ArticleGroups):
			mask[i] = np.random.randint(2, size = self.dimension)
		return mask

	def simulateArticlePool(self):
		articles = []

		articles_id = {}
		mask = self.generateMasks()

		for i in range(self.ArticleGroups):
			articles_id[i] = range((self.n_articles*i)/self.ArticleGroups, (self.n_articles*(i+1))/self.ArticleGroups)

			for key in articles_id[i]:
				featureVector = np.multiply(featureUniform(self.dimension, {}), mask[i])
				l2_norm = np.linalg.norm(featureVector, ord =2)
				articles.append(Article(key, featureVector/l2_norm ))


		return articles


	def small_small_Exp(self, theta, beta, small_bound):
		vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
		#vector = self.FeatureFunc(self.dimension, argv = self.argv)
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)
		while np.dot(vector, theta) > small_bound or np.dot(vector, beta) > small_bound or np.linalg.norm(vector, ord =2) >10:
			#print vector, np.dot(vector, theta), np.dot(vector, beta)
			vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
			vector_l2_norm = np.linalg.norm(vector, ord =2)

		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)

		print 'small, small', np.exp(np.dot(vector, theta)), np.exp(np.dot(vector, beta))
		return vector

	def large_large_Exp(self, theta, beta, large_bound):
		vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)
		while np.dot(vector, theta) < large_bound or np.dot(vector, beta) < large_bound or np.linalg.norm(vector, ord =2) > 10:
			vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
			vector_l2_norm = np.linalg.norm(vector, ord =2)
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)

		print 'lagrge, large', np.exp(np.dot(final_vector_norm, theta)), np.exp(np.dot(final_vector_norm, beta))
		return vector

	def small_large_Exp(self, theta, beta, small_bound, large_bound):
		vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)
		while np.dot(vector, theta) > large_bound or np.dot(vector, beta) < large_bound or np.linalg.norm(vector, ord =2) >10:
			#print vector, np.dot(vector, theta), np.dot(vector, beta)
			vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
			vector_l2_norm = np.linalg.norm(vector, ord =2)
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)

		print 'small, large', np.exp(np.dot(final_vector_norm, theta)), np.exp(np.dot(final_vector_norm, beta))
		return vector

	def large_small_Exp(self, theta, beta, small_bound, large_bound):
		vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)
		while np.dot(vector, theta) < large_bound or np.dot(vector, beta) > large_bound or vector_l2_norm >10:
			vector = np.array([2*np.random.uniform(-1,1) for _ in range(self.dimension)])
			vector_l2_norm = np.linalg.norm(vector, ord =2)
		vector_l2_norm = np.linalg.norm(vector, ord =2)
		final_vector_norm = np.asarray(vector)/float(vector_l2_norm)

		print 'lagrge, small', np.exp(np.dot(final_vector_norm, theta)), np.exp(np.dot(final_vector_norm, beta))
		return vector

	def simulateArticlePool_2SetOfFeature(self):
		articlesDic = {}
		articles_id = {}
		centroids = [0.2, 0.9]

		articlesDic['small_small'] = []
		articlesDic['small_large'] = []
		articlesDic['large_small'] = []
		articlesDic['large_large'] = []
		small_bound = -0.5
		large_bound = 1
		# small_bound_click  = -1   #coressponds to 1/(1+e), which is smaller than 1/2
		# large_bound_click = 1     #coressponds to 1/(1+e^{-1}), which is larger than 1/2

		# small_bound_return = -1  # lambda = -1 coressponds to lambda = e^{-1}, expected return time t = e
		# large_bound_return = 1   # lambda = -1 coressponds to lambda = e^{-1}, expected return time t = 1/e
		print 'generating articles ...'
		for i in range(self.n_articles):
			small_theta_small_beta = list(self.small_small_Exp(self.userFeature_theta, self.userFeature_beta, small_bound))
			small_theta_large_beta = list(self.small_large_Exp(self.userFeature_theta, self.userFeature_beta, small_bound, large_bound))
			large_theta_small_beta = list(self.large_small_Exp(self.userFeature_theta, self.userFeature_beta, small_bound, large_bound))
			large_theta_large_beta = list(self.large_large_Exp(self.userFeature_theta, self.userFeature_beta, large_bound))

			articlesDic['small_small'].append(Article(4*i, 'smallTheta_smallBeta', small_theta_small_beta))
			articlesDic['small_large'].append(Article(4*i+1, 'smallTheta_largeBeta',small_theta_large_beta))
			articlesDic['large_small'].append(Article(4*i+2, 'largeTheta_smallBeta', large_theta_small_beta))
			articlesDic['large_large'].append(Article(4*i+3, 'largeTheta_largeBeta', large_theta_large_beta))
		print 'finish generating articles!'

		return articlesDic['small_small'], 	articlesDic['small_large'], articlesDic['large_small'], articlesDic['large_large']