Cambridge SMT System
data-main.disambig.hpp
Go to the documentation of this file.
1 // Licensed under the Apache License, Version 2.0 (the "License");
2 // you may not use these files except in compliance with the License.
3 // You may obtain a copy of the License at
4 //
5 // http://www.apache.org/licenses/LICENSE-2.0
6 //
7 // Unless required by applicable law or agreed to in writing, software
8 // distributed under the License is distributed on an "AS IS" BASIS,
9 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
10 // See the License for the specific language governing permissions and
11 // limitations under the License.
12 
13 // Copyright 2012 - Gonzalo Iglesias, AdriĆ  de Gispert, William Byrne
14 
15 #ifndef DATA_MAIN_APPLYLM_HPP
16 #define DATA_MAIN_APPLYLM_HPP
17 
25 namespace ucam {
26 namespace fsttools {
32 struct DisambigData {
34  sidx ( 0 ),
35  recasingvcblm ( NULL ),
36  stats ( new StatsData ) {
37  };
38 
40  uint sidx;
41 
43  unordered_map<std::string, std::vector <const KenLMData*> > klm;
45  unordered_set<std::string> *recasingvcblm;
47  unordered_map<std::string, void * > fsts;
48 
50 
52  unordered_map<std::string, ucam::util::WordMapper *> wm;
53 
54 };
55 
56 }
57 } // end namespaces
58 #endif
Contains data for statistics, i.e. allows timing actions and methods called during execution...
Definition: data.stats.hpp:88
data structure for disambig tool. It needs at least a mixed case language model, mixed-case vocabular...
unordered_map< std::string, std::vector< const KenLMData * > > klm
lists of language models indexed by a key (i.e. parameter )
unordered_set< std::string > * recasingvcblm
Mixed case vocabulary used by the unigram model and the language model.
unordered_map< std::string, void * > fsts
Unigram model to be stored here.
unordered_map< std::string, ucam::util::WordMapper * > wm
Wordmap/Integer map objects.
Definition: bleu.hpp:14