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
|
/*
Copyright (c) 1998 - 2024
ILK - Tilburg University
CLST - Radboud University
CLiPS - University of Antwerp
This file is part of timbl
timbl is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
timbl is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, see <http://www.gnu.org/licenses/>.
For questions and suggestions, see:
https://github.com/LanguageMachines/timbl/issues
or send mail to:
lamasoftware (at ) science.ru.nl
*/
#ifndef TIMBL_FEATURES_H
#define TIMBL_FEATURES_H
#include <vector>
#include <map>
#include <unordered_map>
#include "timbl/MsgClass.h"
#include "timbl/Matrices.h"
#include "ticcutils/Unicode.h"
namespace Hash {
class UnicodeHash;
}
namespace Timbl {
class ValueClass;
class TargetValue;
class Targets;
class metricClass;
class SparseValueProbClass {
friend std::ostream& operator<< ( std::ostream&, SparseValueProbClass * );
public:
using IDmaptype = std::map< size_t, double >;
using IDiterator = IDmaptype::const_iterator;
explicit SparseValueProbClass( size_t d ): dimension(d) {};
void Assign( const size_t i, const double d ) { vc_map[i] = d; };
void Clear() { vc_map.clear(); };
IDiterator begin() const { return vc_map.begin(); };
IDiterator end() const { return vc_map.end(); };
private:
IDmaptype vc_map;
size_t dimension;
};
enum FeatVal_Stat {
Unknown,
Singleton,
SingletonNumeric,
NumericValue,
NotNumeric
};
class FeatureValue: public ValueClass {
friend class Feature;
friend class Feature_List;
friend struct D_D;
public:
explicit FeatureValue( const icu::UnicodeString& );
FeatureValue( const icu::UnicodeString&, size_t );
~FeatureValue() override;
void ReconstructDistribution( const ClassDistribution& vd ) {
TargetDist.Merge( vd );
_frequency = TargetDist.totalSize();
};
bool isUnknown() const { return _index == 0; };
SparseValueProbClass *valueClassProb() const { return ValueClassProb; };
private:
SparseValueProbClass *ValueClassProb;
ClassDistribution TargetDist;
};
class Feature: public MsgClass {
friend class MBLClass;
friend class Feature_List;
public:
explicit Feature( Hash::UnicodeHash *T );
~Feature() override;
bool Ignore() const { return ignore; };
void Ignore( const bool val ){ ignore = val; };
bool setMetricType( const MetricType );
MetricType getMetricType() const;
double Weight() const { return weight; };
void SetWeight( const double w ) { weight = w; };
double InfoGain() const { return info_gain; };
void InfoGain( const double w ){ info_gain = w; };
double SplitInfo() const { return split_info; };
void SplitInfo( const double w ){ split_info = w; };
double GainRatio() const { return gain_ratio; };
void GainRatio( const double w ){ gain_ratio = w; };
double ChiSquare() const { return chi_square; };
void ChiSquare( const double w ){ chi_square = w; };
double SharedVariance() const { return shared_variance; };
void SharedVariance( const double w ){ shared_variance = w; };
double StandardDeviation() const { return standard_deviation; };
void StandardDeviation( const double w ){ standard_deviation = w; };
double Min() const { return n_min; };
void Min( const double val ){ n_min = val; };
double Max() const { return n_max; };
void Max( const double val ){ n_max = val; };
double fvDistance( const FeatureValue *,
const FeatureValue *,
size_t=1 ) const;
FeatureValue *add_value( const icu::UnicodeString&, TargetValue *, int=1 );
FeatureValue *add_value( size_t, TargetValue *, int=1 );
FeatureValue *Lookup( const icu::UnicodeString& ) const;
bool decrement_value( FeatureValue *, const TargetValue * );
bool increment_value( FeatureValue *, const TargetValue * );
size_t EffectiveValues() const;
size_t TotalValues() const;
bool isNumerical() const;
bool isStorableMetric() const;
bool AllocSparseArrays( size_t );
void InitSparseArrays();
bool ArrayRead(){ return vcpb_read; };
bool matrixPresent( bool& ) const;
size_t matrix_byte_size() const;
bool store_matrix( int = 1 );
void clear_matrix();
bool fill_matrix( std::istream& );
void print_matrix( std::ostream&, bool = false ) const;
void print_vc_pb_array( std::ostream& ) const;
bool read_vc_pb_array( std::istream & );
FeatVal_Stat prepare_numeric_stats();
void Statistics( double, const Targets&, bool );
void NumStatistics( double, const Targets&, int, bool );
void ClipFreq( size_t f ){ matrix_clip_freq = f; };
size_t ClipFreq() const { return matrix_clip_freq; };
SparseSymetricMatrix<const ValueClass *> *metric_matrix;
private:
Feature( const Feature& );
Feature& operator=( const Feature& );
Hash::UnicodeHash *TokenTree;
metricClass *metric;
bool ignore;
bool numeric;
bool vcpb_read;
enum ps_stat{ ps_undef, ps_failed, ps_ok, ps_read };
enum ps_stat PrestoreStatus;
MetricType Prestored_metric;
void delete_matrix();
double entropy;
double info_gain;
double split_info;
double gain_ratio;
double chi_square;
double shared_variance;
double standard_deviation;
size_t matrix_clip_freq;
std::vector<long int> n_dot_j;
std::vector<long int> n_i_dot;
double n_min;
double n_max;
double weight;
void Statistics( double );
void NumStatistics( std::vector<FeatureValue *>&, double );
void ChiSquareStatistics( const std::vector<FeatureValue *>&,
const Targets& );
void ChiSquareStatistics( const Targets& );
void SharedVarianceStatistics( const Targets&, int );
void StandardDeviationStatistics();
std::vector<FeatureValue *> values_array;
std::unordered_map< size_t, FeatureValue *> reverse_values;
bool is_reference;
};
class Feature_List: public MsgClass {
friend class MBLClass;
public:
Feature_List():
_eff_feats(0),
_num_of_feats(0),
_num_of_num_feats(0),
_feature_hash(0),
_is_reference(false)
{
}
explicit Feature_List( Hash::UnicodeHash *hash ):
Feature_List()
{
_feature_hash = hash;
}
Feature_List &operator=( const Feature_List& );
~Feature_List() override;
void init( size_t, const std::vector<MetricType>& );
Hash::UnicodeHash *hash() const { return _feature_hash; };
size_t effective_feats(){ return _eff_feats; };
Feature *operator[]( size_t i ) const { return feats[i]; };
void write_permutation( std::ostream & ) const;
void calculate_permutation( const std::vector<double>& );
size_t _eff_feats;
size_t _num_of_feats;
size_t _num_of_num_feats;
std::vector<Feature *> feats;
std::vector<Feature *> perm_feats;
std::vector<size_t> permutation;
private:
Hash::UnicodeHash *_feature_hash;
bool _is_reference;
};
} // namespace Timbl
#endif // TIMBL_FEATURES_H
|