Can I read Chinese Textual Entailment Recognition Model Based on Contextual Feature Extraction on EtoBox?

Chinese Textual Entailment Recognition Model Based on Contextual Feature Extraction by Nanhui Hu; Lirong Sui is a scholarly article available to read on EtoBox.

What is Chinese Textual Entailment Recognition Model Based on Contextual Feature Extraction about?

Textual entailment recognition is widely used in machine translation, reading comprehension and other fields. Aiming at the problem that the existing textual entailment recognition models fail to make full use of textual contextual semantic information, this paper constructs a Chinese textual entailment recognition model based on contextual feature extraction. The model consists of input layer, embedding layer, encoding layer, interaction layer, prediction layer and output layer. Among them, the encoding layer adopts a dualencoding network for text. Firstly, Sentence-State LSTM is used to encode the text to obtain rich word-level contextual features, and then self-attention mechanism is added to enhance the ability of text to obtain textual contextual features from a long distance; the interaction layer designs an enhanced contextual cross-attention mechanism, which takes the hidden states of the text sequence encoded by Sentence-State LSTM as the contextual representation and connects it with the text sequence output by the encoding layer, so as to make full use of the contextual information of the text and achieve better text alignment. This paper evaluates the model on two datas

Author

Nanhui Hu; Lirong Sui

Publisher

ACM

Published

2022

Language

EN