P-CNN: Enhancing text matching with positional convolutional neural network

作者：

Highlights：

• Classify text positional information into multiple perspectives.

• Incorporate positional information with interaction signals for text matching.

• Propose a position-sensible convolution filter for the convolutional neural network.

• Empirical insights into the effect of positional information on text matching tasks.

摘要

•Classify text positional information into multiple perspectives.•Incorporate positional information with interaction signals for text matching.•Propose a position-sensible convolution filter for the convolutional neural network.•Empirical insights into the effect of positional information on text matching tasks.

论文关键词：Position,Convolutional neural network,Text matching,Web search,Answer selection

论文评审过程：Received 5 July 2018, Revised 22 January 2019, Accepted 27 January 2019, Available online 2 February 2019, Version of Record 18 February 2019.

论文官网地址：https://doi.org/10.1016/j.knosys.2019.01.028