Finding Function in Form: Compositional Character Models for Open Vocabulary Word Representation

EMNLP 2015 Wang LingTiago LuísLuís MarujoRamón Fernandez AstudilloSilvio AmirChris DyerAlan W. BlackIsabel Trancoso

We introduce a model for constructing vector representations of words by composing characters using bidirectional LSTMs. Relative to traditional word representation models that have independent vectors for each word type, our model requires only a single vector per character type and a fixed set of parameters for the compositional model... (read more)

PDF Abstract

Evaluation results from the paper

Task Dataset Model Metric name Metric value Global rank Compare
Part-Of-Speech Tagging Penn Treebank Char Bi-LSTM Accuracy 97.78 # 3
Part-Of-Speech Tagging Penn Treebank Bi-LSTM Accuracy 97.36 # 10