Seq2Emoji: A hybrid sequence generation model for short text emoji prediction

作者：

Highlights：

• Seq2Emoji, a hybrid sequence generation model, is proposed for predicting emojis matching short texts.

• The model can learn the semantics between the text and emoji labels, as well as the correlation between predicted emojis.

• The model performs well in the diversity of emojis with the diverse beam search algorithm.

• Our model has good performance on the emoji prediction task.

摘要

•Seq2Emoji, a hybrid sequence generation model, is proposed for predicting emojis matching short texts.•The model can learn the semantics between the text and emoji labels, as well as the correlation between predicted emojis.•The model performs well in the diversity of emojis with the diverse beam search algorithm.•Our model has good performance on the emoji prediction task.

论文关键词：00-01,99-00,Emoji prediction,Multi-label classification,Social media,Neural network

论文评审过程：Received 27 July 2020, Revised 24 December 2020, Accepted 28 December 2020, Available online 9 January 2021, Version of Record 12 January 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106727