Research on Sentiment Analysis Based on Short Video Data

Jiayi She

doi:10.47297/wspciWSP2516-252724.20261003

PDF
Export
Share
Collection
Album

Research on Sentiment Analysis Based on Short Video Data

Volume 10, Issue 3

Views: 0 Downloads: 0

Research on Sentiment Analysis Based on Short Video Data
Research on Sentiment Analysis Based on Short Video Data
Guangdong Polytechnic Normal University, Guangzhou, Guangdong, 510450, China
Vol. 10, Issue 3, Pages: 117-122(2026)
Online First：07 August 2026，

Published：30 June 2026
DOI：10.47297/wspciWSP2516-252724.20261003
Accepted：

Scan QR Code

She Jiayi.Research on Sentiment Analysis Based on Short Video Data[J].Creativity and Innovation,2026,10(03):117-122.
DOI：

She Jiayi.Research on Sentiment Analysis Based on Short Video Data[J].Creativity and Innovation,2026,10(03):117-122. DOI： 10.47297/wspciWSP2516-252724.20261003.

摘要

Abstract

Comments on short videos are predominantly everyday spoken language. In different contexts

the same expression carries distinct meanings

which should be interpreted according to contextual information. Furthermore

internet slang emerges frequently

and ironic expressions are widely prevalent. These characteristics significantly raise the difficulty of sentiment analysis. This paper systematically reviews the core literature in this field from 2020 to 2025. Integrating literature induction with content analysis

it establishes a three-dimensional research framework involving technology

application scenarios and theory. Furthermore

it summarizes the technological evolution trajectories

research hotspots

and practical application challenges. The results show that the evolution of sentiment analysis for short video comments can be divided into the following stages: traditional machine learning

deep learning

pre-trained models

lightweight models

and multimodal fusion. Mainstream models have achieved remarkable improvements in accuracy. According to existing studies

the state-of-the-art accuracy in Chinese irony recognition is 78%

and lightweight models can reduce the size of the original BERT model by approximately 70%. Nevertheless

several challenges remain

including inadequate cross-modal fusion

the absence of dynamic adaptation mechanisms for new internet vocabulary

underdeveloped lightweight deployment solutions

and significant shortcomings in data ethics and compliance. his study suggests future research directions including cross-cultural adaptation

internet neologism handling

causal inference

and engineering deployment

providing technical references for content moderation

personalized recommendations

and brand public opinion monitoring on short video platforms.

关键词

Keywords

references

Wang, X.X. (2025). Research on the current status of domestic and foreign short video research based on bibliometric analysis.

Zhong, Z.H., Xiao, J.H., Wu, Y., & Wang, X.C. (2021). Research on popularity modeling of on line short videos based on Douyin platform. J. Electron. Sci. Tech. Univ. , 50(5): 774–781.

Wu, J.Y., & Zhang, P.P. (2022). Research on emotional turn in comments of vaccination short videos. News Knowl. , (11): 78–88.

Liu, M.P. (n.d.) Research on key technologies of video emotion analysis. Doctoral dissertation, Southeast University.

Li, L. (2024). Research on sentiment analysis of online consumer reviews based on BERT: A case study of mobile phone products. Doctoral dissertation, Hangzhou Dianzi University.

Wei, S.X. (n.d.). Research on short video sentiment analysis and irony recognition based on deep learning. Doctoral dissertation, Lanzhou Univ. Technol.

Hong, J.J. (n.d.). Multimodal sentiment analysis based on audio and video enhanced text. Doctoral dissertation, Yangtze University.

Kang, C.Y., & Pei, M.F. (2025). Visual analysis of research context and hotspots of short video emotional communication. Commun. Copyr. , (6).

Chen, J.H., Wang, Y.F., & Nie, H.Z. (n.d.). Research on analysis model of public opinion evolution of short videos in sudden events. J. Inf. Resour. Manage.

Liu, T.L., Xiao, Y.H., & Dai, X.B. (2022). Method and system for annotation associated short video emotion recognition based on vision and language. China Patent No. CN114882412A.

Yang, J.T. (n.d.). Research on key technologies of sentiment analysis based on multimodal fusion. Doctoral dissertation, University Name.

Wang, X.B. (n.d.). Text sentiment analysis of online public opinion events based on multi-source data: A case study of China-US trade war. Doctoral dissertation, Xiangtan University.

Xie, F. (2023). Research on short text sentiment analysis of social network platforms based on deep learning. Doctoral dissertation, Nanchang University.

Zhang, Y.M., & Li, Y. (2026). Fine-grained sentiment analysis of microblog texts based on BERT model. Mod. Inf. Technol. , 10(4): 112–115.

Chan, Y.J. (2025). Survey of sentiment analysis based on deep learning. Inf. Comput. , (21): 78–80.

Liu, X.W. (n.d.). Research on video sentiment analysis based on multimodal fusion. Doctoral dissertation, Nanchang University.

Zhou, J., Ye, J.M., & Li, C. (2021). Multimodal learning emotional computing: Motivation, framework, and suggestions. e-Educ. Res. , 42(7): 8.

Shi, J.Y. (n.d.). Study on influencing factors of emotional communication effect of short videos of mainstream media: Taking Tiktok short videos of the Central Committee of the Communist Youth League as an example. Doctoral dissertation, Yangzhou University.

Zhao, X.T., & Luo, N. (2024). Analysis of user demand characteristics of cultural short videos: Based on KANO model. J. News Res. , 15(6): 17–20.

Song, X. (2022). Research on emotional turn in short video content production of mainstream media. Doctoral dissertation, Hunan University.