Repository navigation
Expand file tree
/
Copy pathModel.cs
More file actions
313 lines (267 loc) · 9.58 KB
/
Copy pathModel.cs
File metadata and controls
313 lines (267 loc) · 9.58 KB
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
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using System.IO;
using Bigtree.Algorithm.CRFLite.Utils;
namespace Txt2Vec
{
public class Term
{
public string strTerm;
public float[] vector;
public byte[] vectorVQ;
}
public class Model
{
private Dictionary<string, Term> term2vector;
private List<Term> entireTermList;
private int vectorSize;
private double[][] codebooks;
public List<Term> Vocabulary { get { return entireTermList; } }
public int VectorSize { get { return vectorSize; } }
public string[] GetAllTerms()
{
return term2vector.Keys.ToArray();
}
public Term GetTerm(string strTerm)
{
if (term2vector.ContainsKey(strTerm) == false)
{
return null;
}
return term2vector[strTerm];
}
public void LoadModel(string strFileName, bool bTextFormat)
{
if (bTextFormat == true)
{
LoadTextModel(strFileName);
}
else
{
LoadBinaryModel(strFileName);
}
}
public bool DumpModel(string strFileName)
{
if (entireTermList == null || entireTermList.Count == 0)
{
return false;
}
StreamWriter sw = new StreamWriter(strFileName);
foreach (Term term in entireTermList)
{
StringBuilder sb = new StringBuilder();
sb.Append(term.strTerm);
sb.Append("\t");
foreach (double v in term.vector)
{
sb.Append(v);
sb.Append("\t");
}
sw.WriteLine(sb.ToString().Trim());
}
sw.Close();
return true;
}
public void LoadTextModel(string strFileName)
{
term2vector = new Dictionary<string, Term>();
entireTermList = new List<Term>();
vectorSize = 0;
StreamReader sr = new StreamReader(strFileName);
string strLine = null;
while ((strLine = sr.ReadLine()) != null)
{
//the format is "word \t vector
//eah dim of vector is splitted by \t
Term term = new Term();
string[] items = strLine.Split('\t');
int vSize = items.Length - 1;
if (vectorSize > 0 && vectorSize != vSize)
{
throw new InvalidDataException(String.Format("Invalidated data : {0} . The length of vector must be fixed (current length {1} != previous length {2}).", strLine, vSize, vectorSize));
}
term.strTerm = items[0];
term.vector = new float[vSize];
for (int i = 0; i < vSize; i++)
{
term.vector[i] = float.Parse(items[i + 1]);
}
vectorSize = vSize;
term.vector = NormalizeVector(term.vector);
term2vector.Add(term.strTerm, term);
entireTermList.Add(term);
}
sr.Close();
}
public float[] GetVector(string strTerm)
{
if (term2vector.ContainsKey(strTerm) == true)
{
return term2vector[strTerm].vector;
}
return null;
}
private float[] NormalizeVector(float[] vec)
{
//Normalize the vector
double len = 0;
for (int a = 0; a < vectorSize; a++)
{
len += vec[a] * vec[a];
}
len = Math.Sqrt(len);
for (int a = 0; a < vectorSize; a++)
{
vec[a] = (float)(vec[a] / len);
}
return vec;
}
public void LoadBinaryModel(string strFileName)
{
StreamReader sr = new StreamReader(strFileName);
BinaryReader br = new BinaryReader(sr.BaseStream);
//The number of words
int words = br.ReadInt32();
//The size of vector
vectorSize = br.ReadInt32();
int vqSize = br.ReadInt32();
term2vector = new Dictionary<string, Term>();
entireTermList = new List<Term>();
// Logger.WriteLine("vocabulary size: {0}, vector size: {1}, VQ size: {2}", words, vectorSize, vqSize);
codebooks = null;
if (vqSize > 0)
{
//Read code books
codebooks = new double[vectorSize][];
for (int i = 0; i < vectorSize; i++)
{
codebooks[i] = new double[vqSize];
for (int j = 0; j < vqSize; j++)
{
codebooks[i][j] = br.ReadDouble();
}
}
}
for (int b = 0; b < words; b++)
{
Term term = new Term();
term.strTerm = br.ReadString();
term.vector = new float[vectorSize];
if (codebooks != null)
{
term.vectorVQ = new byte[vectorSize];
}
else
{
term.vectorVQ = null;
}
for (int i = 0; i < vectorSize; i++)
{
if (codebooks == null)
{
term.vector[i] = br.ReadSingle();
}
else
{
byte idx = br.ReadByte();
term.vector[i] = (float)codebooks[i][idx];
term.vectorVQ[i] = idx;
}
}
term.vector = NormalizeVector(term.vector);
term2vector.Add(term.strTerm, term);
entireTermList.Add(term);
}
sr.Close();
}
public static void SaveModel(string strFileName, int vocab_size, int vector_size, List<vocab_word> vocab, double[] syn)
{
StreamWriter fo = new StreamWriter(strFileName);
BinaryWriter bw = new BinaryWriter(fo.BaseStream);
// Logger.WriteLine("Saving term and vector into model file...");
// Save the word vectors
bw.Write(vocab_size);
bw.Write(vector_size);
bw.Write(0); //no VQ
for (int i = 0; i < vocab_size; i++)
{
//term string
bw.Write(vocab[i].word);
//term vector
for (int j = 0; j < vector_size; j++)
{
bw.Write((float)(syn[i * vector_size + j]));
}
}
bw.Flush();
fo.Flush();
fo.Close();
}
public bool BuildVQModel(string strFileName)
{
int vqSize = 256;
if (entireTermList == null || entireTermList.Count == 0)
{
return false;
}
StreamWriter fo = new StreamWriter(strFileName);
BinaryWriter bw = new BinaryWriter(fo.BaseStream);
// Save the word vectors
bw.Write(entireTermList.Count); //Vocabulary size
bw.Write(vectorSize); //Vector size
bw.Write(vqSize); //VQ size
// Logger.WriteLine("vocabulary size: {0}, vector size: {1}, vq size: {2}", entireTermList.Count, vectorSize, vqSize);
//Create word and VQ values mapping table
Dictionary<string, List<byte>> vqResult = new Dictionary<string, List<byte>>();
foreach (Term term in entireTermList)
{
vqResult.Add(term.strTerm, new List<byte>());
}
// Logger.WriteLine("Dims Distortion:");
for (int i = 0; i < vectorSize; i++)
{
//Generate VQ values for each dimension
VectorQuantization vq = new VectorQuantization();
for (int j = 0; j < entireTermList.Count; j++)
{
vq.Add(entireTermList[j].vector[i]);
}
double distortion = vq.BuildCodebook(vqSize);
// Logger.WriteLine("Dim {0}: {1}", i, distortion);
for (int j = 0; j < entireTermList.Count; j++)
{
byte vqValue = (byte)vq.ComputeVQ(entireTermList[j].vector[i]);
vqResult[entireTermList[j].strTerm].Add(vqValue);
}
//Save VQ codebook into model file
for (int j = 0; j < vqSize; j++)
{
bw.Write(vq.CodeBook[j]);
}
}
foreach (KeyValuePair<string, List<byte>> pair in vqResult)
{
if (pair.Value.Count != vectorSize)
{
throw new Exception(String.Format("word {0} has inconsistent vector size: orginial size is {1}, vq size is {2}",
pair.Key, vectorSize, pair.Value.Count));
}
//term string
bw.Write(pair.Key);
//term vector
for (int b = 0; b < pair.Value.Count; b++)
{
bw.Write(pair.Value[b]);
}
}
bw.Flush();
fo.Flush();
fo.Close();
return true;
}
}
}