| J00-3003 | . First , as we saw earlier , | word-based classification | suffers from recognition errors |
| P10-2062 | OpinionFinder models the task as a | word-based classification | problem rather than a sequence |
| W15-1815 | for the informal texts . While | word-based classification | performs well for the indomain |
| N03-1001 | 1-best recognition output ) when a | word-based classification | algorithm ( the one described |
| J14-2008 | annotations are formulated as a | word-based classification | problem , where each word in |
| N01-1031 | accuracies as high or higher than | word-based classification | while avoiding the risk of overtraining |
