blob: 6c1660fabbb5099d1d30f84ab1142fd908355995 (
plain)
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
|
#ifndef MERT_BLEU_SCORER_H_
#define MERT_BLEU_SCORER_H_
#include <iostream>
#include <string>
#include <vector>
#include "Types.h"
#include "ScoreData.h"
#include "Scorer.h"
#include "ScopedVector.h"
using namespace std;
/**
* Bleu scoring
*/
class BleuScorer: public StatisticsBasedScorer
{
public:
explicit BleuScorer(const string& config = "");
~BleuScorer();
virtual void setReferenceFiles(const vector<string>& referenceFiles);
virtual void prepareStats(size_t sid, const string& text, ScoreStats& entry);
virtual float calculateScore(const vector<int>& comps) const;
virtual size_t NumberOfScores() const { return 2 * kLENGTH + 1; }
private:
enum ReferenceLengthType {
AVERAGE,
SHORTEST,
CLOSEST
};
/**
* A NgramCounts is a key-value store.
* Clients don't have to worry about the actual implementation
* since this type is used in internal only.
*/
class NgramCounts;
/**
* Count the ngrams of each type, up to the given length in the input line.
*/
size_t countNgrams(const string& line, NgramCounts& counts, unsigned int n);
void dump_counts(const NgramCounts& counts) const;
// For calculating effective reference length.
void CalcAverage(size_t sentence_id,
vector<ScoreStatsType>& stats) const;
void CalcClosest(size_t sentence_id, size_t length,
vector<ScoreStatsType>& stats) const;
void CalcShortest(size_t sentence_id,
vector<ScoreStatsType>& stats) const;
const int kLENGTH;
ReferenceLengthType m_ref_length_type;
// data extracted from reference files
ScopedVector<NgramCounts> m_ref_counts;
vector<vector<size_t> > m_ref_lengths;
// no copying allowed
BleuScorer(const BleuScorer&);
BleuScorer& operator=(const BleuScorer&);
};
#endif // MERT_BLEU_SCORER_H_
|