Chinese Short Text Classification by ERNIE Based on LTC_Block

Short text classification, an important direction of the basic research natural language processing, has extensive applications. Its effect depends on feature extraction methods and representation methods. This paper proposed LTC_Block-based short classification model named ERNIE to classify Chinese texts extract semantics in corpus address polysemy problem text. In this model, LTC_Block, a double-channel structural unit composed BiLSTM TextCNN, was used contextual sequences overall features semantics, residual connection integrate further texts. Experiments two different datasets showed that achieved better than mainstream models, proving its feasibility effectiveness.