The research results of Tianyi Cloud AI team were included in the top international conference ACL 2023

Recently, the latest research results of the AI team of Tianyi Cloud Intelligent Edge Division "Emotion classification on code-mixed text messages via soft prompt tuning" were included in the ACL 2023 WASSA Workshop.

The ACL (Annual Meeting of the Association for Computational Linguistics) conference is the top international conference in the field of natural language processing. It is also a Class A international academic conference recommended by the China Computer Federation (CCF, China Computer Federation).

Due to the existence of mixed language and non-character information (such as emoticons, etc.), emotion classification of general scene text information has always been a challenging problem. In order to deal with the above difficulties, Tianyi Cloud's artificial intelligence team proposed a soft prompt model tuning method. This method is lightweight and efficient, and it can deeply explore the potential capabilities of pre-trained language models, thereby effectively improving the accuracy of sentiment classification. This paper adopts a novel pre-processing method to process emoticons to better exploit the rich emotional information they contain. In addition, this paper applies a variety of prompt templates (templates) and answer space mapping (verbalizers) to improve the emotion classification effect. A large number of experimental results show that the use of emoticon preprocessing and prompt tuning are beneficial to improving classification performance.

Artificial intelligence is a strategic technology that leads a new round of scientific and technological revolution and industrial transformation. It is also an important starting point for the development of the digital economy. It is of great significance to building a network power and a digital China.

The above is the detailed content of The research results of Tianyi Cloud AI team were included in the top international conference ACL 2023. For more information, please follow other related articles on the PHP Chinese website!