Author(s)

Cheng Zhao^1,2

Affiliation(s)

¹ Command and Control Engineering College, Army Engineering University of PLA, Nanjing 212000, Jiangsu, China;

² School of Computer and Information, Anhui Normal University, Wuhu 241000, Anhui, China.

Corresponding Author

Cheng Zhao, email: ntmaple@mail.ahnu.edu.cn

ABSTRACT

In order to improve the accuracy of social network users' emotional behavior prediction, this paper proposes a multi-level attention mechanism based microblog user group emotional trend prediction. Based on Sina Weibo data, the multi-level attention mechanism is used to divide the text content emotion of user groups and collect the emotional characteristics of user groups. This paper analyzes the group emotional behavior of microblog users and establishes a hierarchical emotion processing model. The model of emotion transfer of microblog users is constructed, and the problem of unstable convergence value is solved by direct fuzzy algorithm. Using the node matching method to measure the emotional similarity, the sentiment trend of microblog users is predicted. Taking the microblog data from November to December in 2019 as an example, the experimental results show that the emotional trend predicted by this method is close to the actual trend, which shows the effectiveness of the method in this paper.

KEYWORDS

Multi level attention; microblog users; group emotional trend; feature collection.

CITE THIS PAPER

Zhao Cheng. Sentiment trend prediction of microblog users based on multi-level attention mechanism. Eurasia Journal of Science and Technology. 2020, 2(2): 64-70.

REFERENCES

[1] Birjali M, Beni-Hssane A, Erritali M. Machinelearning and semantic sentiment analysis basedalgorithms for suicide sentiment prediction in socialnetworks[J]. Procedia Computer Science, 2017, 113:65-72. 

[2] Hong K , Liu G , Chen W , et al. Classification of theemotional stress and physical stress using signalmagnification and canonical correlation analysis[J]. Pattern Recognition, 2017, 77(6):140-149. 

[3] Li X, Rao Y, Xie H, et al. Bootstrapping socialemotion classification with semantically rich hybridneural networks[J]. IEEE Transactions on AffectiveComputing, 2017, 8(4): 428-442. 

[4] Budiharto W, Meiliana M. Prediction and analysis ofIndonesia Presidential election from Twitter usingsentiment analysis[J]. Journal of Big data, 2018, 5(1): 51. 

[5] Das S, Behera R K, Rath S K. Real-time sentimentanalysis of twitter streaming data for stockprediction[J]. Procedia computer science, 2018, 132:956-964. 

[6] Ghosal D, Akhtar M S, Ekbal A, et al. Deep ensemblemodel with the fusion of character, word andlexicon level information for emotion and sentimentprediction[C]//International Conference on NeuralInformation Processing. Springer, Cham, 2018:162-174. 

[7] Agnès Ros-Morente, Cuenca E C , Guiu G F . Analysis of the Effects of two Gamified EmotionalEducation Software’s in Emotional and Well-beingVariables in Spanish Children and Adolescents[J]. International Journal of Emerging Technologies inLearning (iJET), 2018, 13(9):148. 

[8] Pilato G , D’Avanzo, Ernesto. Data-driven SocialMood Analysis through the Conceptualization ofEmotional Fingerprints[J]. Procedia ComputerScience, 2018, 123(4):360-365. 

[9] Ramadhani A M, Goo H S. Twitter sentiment analysisusing deep learning methods[C]//2017 7thInternational annual engineering seminar (InAES). IEEE, 2017: 1-4. 

[10] Ahmad A , Paul A , Din S , et al. Multilevel DataProcessing Using Parallel Algorithms for AnalyzingBig Data in High-Performance Computing[J]. International journal of parallel programming, 2018, 46(3):508-527. 

[11] Sinha S, Saxena K, Joshi N. Sentiment Prediction ofFacebook Status Updates of Youngsters[M]//Data, Engineering and Applications. Springer, Singapore, 2019: 95-105. 

[12] Naveh S , Bronstein J . Sense making in complexhealth situations: Virtual health communities assources of information and emotional support[J]. Aslib Journal of Information Management, 2019, 71(6):789-805. 

[13] Gutiérrez I, Gómez D, Castro J, et al. A newcommunity detection algorithm based on fuzzymeasures[C]//International Conference onIntelligent and Fuzzy Systems. Springer, Cham, 2019: 133-140.

[14] Nieuwazny J , Masui F , Ptaszynski M , et al. Howreligion and morality correlate in age of society 5.0:Statistical analysis of emotional and moralassociations with Buddhist religious termsappearing on Japanese blogs[J]. Cognitive SystemsResearch, 2020, 59(Jan.):329-344. 

[15] Hong K , Liu G , Chen W , et al. Classification ofthe emotional stress and physical stress using signalmagnification and canonical correlation analysis[J]. Pattern Recognition, 2017, 77(10):140-149.