| C92-2095 | eliminate unnecessary center - | embedding /embedding/NN | ; and ( 3 ) eliminating of scrambling |
| D15-1252 | , penalizing embeddings , re - | embedding /embedding/NN | words , and dropout . We also |
| C80-1074 | Fl ) , nominalization ( FII ) , | embedding /embedding/NN | ( fill ) , connecting ( FIV ) |
| C86-1008 | axiomatic theory of dialogue , | embedding /embed/VBG | rhetorical patterns , focusing |
| P10-1121 | HHMM framework , a new metric , | embedding /embedding/NN | difference , is also proposed |
| P00-1011 | select from a set of noun phrases , | embedding /embed/VBG | proper names of different semantic |
| W14-5201 | pipelines with other researchers , | embedding /embed/VBG | NLP pipelines in applications |
| J82-3001 | different . The intuitive notion of " | embedding /embed/VBG | a linguistic theory into a model |
| C82-2022 | of PP 's , for which both the " | embedding /embedding/NN | " and the " same-level " hypotheses |
| D14-1030 | some learned representation ( | embedding /embedding/NN | ) , and ( 3 ) compute output |
| W15-1511 | 1992 ) or the ( less well-known ) | embedding /embedding/NN | form given by the canonical correlation |
| W13-3207 | introduce a new 50-dimensional | embedding /embedding/NN | obtained by spectral clustering |
| W05-0627 | the largest probability among | embedding /embed/VBG | ones are kept . After predicting |
| J95-2003 | constituent discourse segments ; an | embedding /embedding/NN | relationship may hold between |
| E89-1033 | has been implemented , and an | embedding /embedding/NN | of this in an interactive parsing |
| P15-1107 | that hierarchically builds an | embedding /embedding/NN | for a paragraph from embeddings |
| P15-1104 | the supervised data to find an | embedding /embedding/NN | subspace that fits the task complexity |
| C88-1033 | reformulated as the problem of finding an | embedding /embedding/NN | function f from the representational |
| P98-2242 | given , which are realised in an | embedding /embedding/NN | algorithm . The significant aspect |
| P15-2048 | labels . Specifically , we learn an | embedding /embedding/NN | for each label and each feature |
