Awesome

Multilayer Perceptron Attention

[Embedded in AREkit-0.20.0 and later versions]

UPD December 7rd, 2019: this attention model becomes a part of AREkit framework (original, interactive). Please proceed with this framework for an embedded implementation.

This project is an unofficial implementation of MLP attention -- multilayer perceptron attention network, proposed by Yatian Shen and Xuanjing Huang as an application for Relation Extraction Task [paper].

Vector representation of words and entities includes:

• Term embedding;
• Part-Of-Speech (POS) embedding;
• Distance embedding;

Application and Experiments

You may proceed with the following repositories NLDB-2020 paper/code; WIMS-2020 paper/code.

This version has been embedded in AREkit-[0.20.3], and become a part of the following papers:

• Attention-Based Neural Networks for Sentiment Attitude Extraction using Distant Supervision [ACM-DOI] / [presentation]
  • Rusnachenko Nicolay, Loukachevitch Natalia
  • WIMS-2020
• Studying Attention Models in Sentiment Attitude Extraction Task [Springer] / [arXiv:2006.11605] / [presentaiton]
  • Rusnachenko Nicolay, Loukachevitch Natalia
  • NLDB-2020

References

• Attention-Based Convolutional Neural Network for Semantic Relation Extraction [paper]
  • Yatian Shen and Xuanjing Huang
  • COLING 2016