W07-0710 all subsequent experiments of feature expansion . 4.3 BLEU Training Results We
P14-2043 using a relatively straightforward feature expansion scheme . Experiments on five
P13-1129 alternation pattern by utilizing a feature expansion scheme . For each utterance n
P11-1014 the effect of not performing any feature expansion . We simply train a binary classifier
P11-1014 sentiment sensitive thesaurus for feature expansion . Given a labeled or an unlabeled
P11-1014 Shen et al. , 2009 ) . However , feature expansion techniques have not previously
P14-1130 tensor represents a substantial feature expansion . The arc score stensor ( h --
P11-1014 sentiment sensitive thesaurus for feature expansion is useful for cross-domain sentiment
D15-1049 preceding paragraph : an intuitive feature expansion representation of the domain
W07-0710 have introduced a non-parametric feature expansion , which guarantees invariance
W13-2250 At the end , when all possible feature expansions are considered , each example
P11-1014 reviews . 4 Feature Expansion Our feature expansion phase augments a feature vector
P11-1014 from a large set of reviews . 4 Feature Expansion Our feature expansion phase augments
P07-2017 the testing speed of different feature expansion techniques , namely , array visiting
W14-3411 using a generic solution such as feature expansion based on distributional similarity
W06-1208 , and were most effective as a feature expansion algorithm . The only obvious
P07-2017 provide a more efficient solution to feature expansion when d is set more than two .
P06-2087 analysis , we propose to base our feature expansion also on argumentative cri - teria
P14-1130 counters the otherwise uncontrolled feature expansion . More - over , by controlling
P03-1004 polynomial kernel allows such feature expansion without loss of generality or
hide detail