| D12-1084 | level is mainly based on phrase | pair extraction | techniques from the MT literature |
| D12-1058 | filtering methods . The phrase | pair extraction | process of phrasebased SMT systems |
| D15-1119 | ( line 11 ) are general phrase | pair extraction | process . The if ( line 13 ) |
| E14-1073 | : 1 . Main ( trigger , theme ) | pair extraction | : main event extraction step |
| D12-1003 | each graph , ( 3 ) translation | pair extraction | . 3.1 Graph Construction We construct |
| D09-1113 | because they have the highest | pair extraction | confidence values . At the same |
| I05-2030 | information assists attributevalue | pair extraction | . Focusing on the rows labeled |
| E09-3004 | similarity . 4.1 Step 1 - Pattern-Based | Pair Extraction | The first step of our algorithm |
| I05-2030 | follows , we discuss the results of | pair extraction | and opinionhood determination |
| E14-1004 | ) yields a conservative phrase | pair extraction | method that reduces phrase-table |
| C04-1151 | precision of parallel sentence | pair extraction | increases steadily over each |
| D08-1033 | function is similar to the phrase | pair extraction | heuristic , but never includes |
| I05-2030 | , we evaluated the results of | pair extraction | and opinionhood determination |
| I05-2030 | Table 2 shows the results . In the | pair extraction | , we assume that the value is |
| E03-1050 | Chinese , was used for phrase | pair extraction | as this resulted in higher translation |
| D12-1003 | experi - ments . 3.3 Translation | Pair Extraction | After label propagations , we |
| E14-1073 | process alike ( trigger , theme ) | pair extraction | , also with constraints avoiding |
| D14-1173 | , allowing for correct phrase | pair extraction | ( Zens et al. , 2002 ) . For |
| D14-1173 | , which hinders correct phrase | pair extractions | . For example , in Figure 2 ( |
| D10-1015 | problems by modifying the phrase - | pair extraction | algorithm so that morpheme-token |
