Tag-Enhanced Dynamic Compositional Neural Network over arbitrary tree structure for sentence representation

作者：

Highlights：

• Learns sentence representation from richer arbitrary tree structures instead of binarized tree structures.

• Dynamic compositional model extended to consider additional structural tag information.

• First to handle arbitrary tree structures and capture the richness of compositionality.

摘要

•Learns sentence representation from richer arbitrary tree structures instead of binarized tree structures.•Dynamic compositional model extended to consider additional structural tag information.•First to handle arbitrary tree structures and capture the richness of compositionality.

论文关键词：Sentence representation,Text classification,Text semantic matching,Dynamic compositional networks

论文评审过程：Received 5 December 2019, Revised 6 April 2021, Accepted 8 May 2021, Available online 14 May 2021, Version of Record 18 May 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115182