Entity-level sentiment prediction in Danmaku video interaction

Bai, Qingchun; Wei, Kai; Zhou, Jie; Xiong, Chao; Wu, Yuanbin; Lin, Xin; He, Liang

doi:10.1007/s11227-021-03652-4

Cited by 10 publications

(8 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep learning-based methods have stronger feature learning capabilities, reducing the cost of building and selecting features [37], but the method needs to be based on a large amount of data, and is prone to data sparsity and overfitting problems in the case of small datasets [38]; Mainstream text pre-training models use different segmentation methods, BERT and XLNet use WordPiece, RoBERTa uses Byte-Pair Encoding, and the use of Chinese segmentation tools can easily lead to the fact that there is no difference between the Chinese segmented corpus and the pre-segmented one after tokenizer processing, which loses semantic information and results in model performance degradation. RQ4: How to effectively extract the word information after Chinese word segmentation when the Chinese word segmentation method is inconsistent with the tokenizer method when using the text pre-training model?…”

Section: Deep Learning-based Danmaku Sentiment Analysismentioning

confidence: 99%

Sentiment Analysis of Video Danmakus Based on MIBE-RoBERTa-FF-BiLSTM

Zhao,

Liu,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

Danmakus are user-generated comments that overlay on videos, enabling real-time interactions between viewers and video content. The emotional orientation of danmakus can reflect the attitudes and opinions of viewers on video segments, which can help video platforms optimize video content recommendation and evaluate users’ abnormal emotion levels. Aiming at the problems of low transferability of traditional sentiment analysis methods in the danmaku domain, low accuracy of danmaku text segmentation, poor consistency of sentiment annotation, and insufficient semantic feature extraction, this paper proposes a video danmaku sentiment analysis method based on MIBE-RoBERTa-FF-BiLSTM. This paper constructs a "Bilibili Must-Watch List and Top Video Danmaku Sentiment Dataset” by ourselves, covering 10,000 positive and negative sentiment danmaku texts of 18 themes. A new word recognition algorithm based on mutual information (MI) and branch entropy (BE) is used to discover 2,610 irregular network popular new words from trigrams to heptagrams in the dataset, forming a domain lexicon. The Maslow’s hierarchy of needs theory is applied to guide the consistent sentiment annotation. The domain lexicon is integrated into the feature fusion layer of the RoBERTa-FF-BiLSTM model to fully learn the semantic features of word information, character information, and context information of danmaku texts and perform sentiment classification. Comparative experiments on the dataset show that the model proposed in this paper has the best comprehensive performance among the mainstream models for video danmaku text sentiment classification, with an F1 value of 94.06%, and its accuracy and robustness are also better than other models. The limitations of this paper are that the construction of the domain lexicon still requires manual participation and review, the semantic information of danmaku video content and the positive case preference are ignored.

show abstract

Section: Deep Learning-based Danmaku Sentiment Analysismentioning

confidence: 99%

Sentiment Analysis of Video Danmakus Based on MIBE-RoBERTa-FF-BiLSTM

Zhao,

Liu,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Deep learning-based methods have stronger feature learning capabilities, reducing the cost of building and selecting features 36 , but the method needs to be based on a large amount of data, and is prone to data sparsity and overfitting problems in the case of small datasets 37 ; Mainstream text pre-training models use different segmentation methods, BERT and XLNet use WordPiece, RoBERTa uses Byte-Pair Encoding, and the use of Chinese segmentation tools can easily lead to the fact that there is no difference between the Chinese segmented corpus and the pre-segmented one after tokenizer processing, which loses semantic information and results in model performance degradation.…”

Section: Related Workmentioning

confidence: 99%

Sentiment analysis of video danmakus based on MIBE-RoBERTa-FF-BiLSTM

Zhao,

Liu,

Wang

et al. 2024

Sci Rep

View full text Add to dashboard Cite

Danmakus are user-generated comments that overlay on videos, enabling real-time interactions between viewers and video content. The emotional orientation of danmakus can reflect the attitudes and opinions of viewers on video segments, which can help video platforms optimize video content recommendation and evaluate users’ abnormal emotion levels. Aiming at the problems of low transferability of traditional sentiment analysis methods in the danmaku domain, low accuracy of danmaku text segmentation, poor consistency of sentiment annotation, and insufficient semantic feature extraction, this paper proposes a video danmaku sentiment analysis method based on MIBE-RoBERTa-FF-BiLSTM. This paper constructs a “Bilibili Must-Watch List and Top Video Danmaku Sentiment Dataset” by ourselves, covering 10,000 positive and negative sentiment danmaku texts of 18 themes. A new word recognition algorithm based on mutual information (MI) and branch entropy (BE) is used to discover 2610 irregular network popular new words from trigrams to heptagrams in the dataset, forming a domain lexicon. The Maslow’s hierarchy of needs theory is applied to guide the consistent sentiment annotation. The domain lexicon is integrated into the feature fusion layer of the RoBERTa-FF-BiLSTM model to fully learn the semantic features of word information, character information, and context information of danmaku texts and perform sentiment classification. Comparative experiments on the dataset show that the model proposed in this paper has the best comprehensive performance among the mainstream models for video danmaku text sentiment classification, with an F1 value of 94.06%, and its accuracy and robustness are also better than other models. The limitations of this paper are that the construction of the domain lexicon still requires manual participation and review, the semantic information of danmaku video content and the positive case preference are ignored.

show abstract

“…Wang et al proposed an improved Bi-LSTM based model to identify the emotions of danmaku messages [11]. Besides, a framework of entity-level sentiment analysis on danmaku video comments was proposed in [12].…”

Section: B Sentiment Analysismentioning

confidence: 99%

Behavior Analytics, Sentiment Analysis, and Topic Detection of Danmaku from Online Electronics Courses on Bilibili

Zeng¹,

Tan²,

Xia³

et al. 2023

IJIET

View full text Add to dashboard Cite

Danmaku data from an online course contains implicit information about the students, the teacher, and the course itself. To discover the information, we design a behavior-sentiment-topic mining procedure, and apply it on the danmaku from two electronics courses on Bilibili, a popular video sharing platform in China. The procedure enables us to obtain behavior patterns, text sentiments, and hidden topics, of those danmaku comments effectively. Results show similarities and differences between the danmaku from Fundamentals of Analog Electronics and that from Fundamentals of Digital Electronics. Some interesting observations are given according to the results. For example, students tend to experience an emotional upsurge right before the end of a course, which is due to their fulfilment for completing the course. Based on the observations, we make some suggestions for students, teachers, and platforms on how to improve the learning outcomes using the results of danmaku analysis.

show abstract

Entity-level sentiment prediction in Danmaku video interaction

Cited by 10 publications

References 32 publications

Sentiment Analysis of Video Danmakus Based on MIBE-RoBERTa-FF-BiLSTM

Sentiment Analysis of Video Danmakus Based on MIBE-RoBERTa-FF-BiLSTM

Sentiment analysis of video danmakus based on MIBE-RoBERTa-FF-BiLSTM

Behavior Analytics, Sentiment Analysis, and Topic Detection of Danmaku from Online Electronics Courses on Bilibili

Contact Info

Product

Resources

About