About YU

News

Professor HWANG Do-sam of Department of Computer Engineering developed “A model to analyze emotion of online commenters. N

No.2335919
  • Writer PR team
  • Date : 2022.04.27 10:12
  • Publication Date : 2022.04.20
  • Views : 6873

Analysis of emotion of Twitter and commenters using deep learning graph technology

Possible use for decision making systems and for searching fake news based on context

Report in recent version of <Information Sciences>, an internationally famous journal in computer science area

[April 20, 2022]

https://www.yu.ac.kr/_attach/yu/editor-image/2022/04/hwpqbmSqnrjrExgRlmbFsXimZF.jpg

<Professor HWANG Do-sam of Department of Computer Engineering (Left) and Research Professor Phan Thi Huyen Trang (Right)>

“Is it possible to evaluate commenters’ emotions through analysis of online comments?”


  Research Team of Professor HWANG Do-sam of YU Department of Computer Engineering showed the answer to this question. Research Team of Professor Hwang developed a new model (CANN-SSCG: Convolutional Attention Neural Network – Semantic Syntactic Context Graph) to analyze emotion of users included in big data. Graphs of structures, meanings, and contexts of written texts were drawn using this model, these three different graphs were integrated to a graph, and emotion of commenters was analyzed using convolutional attention neural network.


  The research team explained the performance of the research saying, “Deep learning and graph technologies in next-generation’s advanced research area were used.When various data including twits and comments were tested, a result improved by 7.75% compared with previous meaning analysis methods was obtained. The developed model may be used for context-based decision-making systems and for searching fake news.”


  The performance of the research was reported in recent version (April, 2022) of < Information Sciences >, an internationally famous journal within higher 11% of computer science area, with the title of “Convolutional attention neural network over graph structures for improving the performance of aspect-level sentiment analysis.”

  The research was jointly conducted with Poland. Research Professor Phan Thi Huyen Trang of YU Department of Computer Engineering participated as lead author and Professor HWANG Do-sam and Professor Ngoc Thanh Nguyen of Wroclaw University of Science and Technology led the research as co-corresponding authors. Especially, the research team attracted attention of academic society with development of a system to help third party’s decision making by analyzing emotion of commenters and measuring satisfaction levels with use of deep learning technology and fuzzy determination structure last year.


  Professor HWANG Do-sam led the research team and said, “Good research performances have been obtained through practical international exchange activities with a Polish research team for many years. The research conducted as BK21 Project of National Research Foundation of Korea is evaluated as a model case of international cooperative joint researches. I plan to develop an emotion analysis system to use an integrated graph structure that includes domain knowledge in addition to structure, meaning, and context graph.”


  Meanwhile, the research was conducted as BK21 Project of National Research Foundation of Korea and Professor HWANG leads credible AI SW system fusion education & research team that uses big data analysis.