Abstract:Bullet screen, a form of online video commentary in emerging social media, is widely used on video websites frequented by young people. It has become a novel means of expressing emotions towards videos. The characteristics, such as varying text lengths and the presence of numerous new words, lead to ambiguous emotional information. To address these characteristics, this paper proposes a Robustly Optimized BERT Pretraining Approach (RoBERTa) + Convolutional Neural Network (CNN) sentiment classification algorithm integrated with a sentiment lexicon. RoBERTa encodes the input text to enhance semantic feature representation, and CNN extracts local features using multiple convolutional kernels of different sizes. Sentiment classification is then performed by a softmax classifier. Meanwhile, we use the sentiment lexicon to calculate the emotion score of the input text and normalize the emotion score. Finally, the classification results of the sentiment lexicon and RoBERTa+CNN are weighted and calculated. The bullet screens are grouped according to their length, and different weights are assigned to the sentiment lexicon based on their length to enhance the features of the model's sentiment classification. The method combines the sentiment lexicon can be customized for the domain vocabulary and the pre-trained model can deal with the polysemy. Experimental results demonstrate that the proposed method achieves improvements in precision, recall, and F1 score. The experiments in this paper take the Russia–Ukraine war as the research topic, and the experimental methods can be extended to other events. The experiment demonstrates the effectiveness of the model in the sentiment analysis of bullet screen texts and has a positive effect on grasping the current public opinion status of hot events and guiding the direction of public opinion in a timely manner.

Natural Language Processing in “bullet Screen” Application

SmartBullets: A Cloud-Assisted Bullet Screen Filter based on Deep Learning

Research on efficient organization and browsing of TV program based on new media platform

The Analysis of Influence Factors of User Experience in Bullet Screen

A Study of Chinese Danmaku Video Sites for Self-directed Learning(CDSDL) from an Educational Perspective: A Case Study of Bilibili Website

BPMiner: mining developers' behavior patterns from screen-captured task videos.

Sentiment Analysis: An ERNIE-BiLSTM Approach to Bullet Screen Comments

A Bullet Screen Sentiment Analysis Method That Integrates the Sentiment Lexicon with RoBERTa-CNN

Bullet Subtitle Sentiment Classification Based on Affective Computing and Ensemble Learning

How to Mine Business Value from Bullet Screen Comments in TV Series by Aspect Sentiment Analysis?

Movie Summarization Using Bullet Screen Comments

Research on the Influence of Uploaders’ Behavior on the Effect of Bullet Screen Video Dissemination

Bullet-screen engagement in videos and deindividualized online behavior: the chain mediating role of belonging and loneliness

Research on Highlight Snippets Identification Technology Based on Sentiment Analysis of Bullet Curtain

Research on sentiment analysis based on the Bilibili video barrage

Chinese Natural Language Processing: From Text Categorization to Machine Translation

Python-based film review data acquisition and visualization design

Response to LiveBot: Generating Live Video Comments Based on Visual and Textual Contexts

It pays to diversify: Effect of bullet‐screen comment diversity on payment

Analysis of the User Behavior and Opinion Classification Based on the BBS

Aspect sentiment mining of short bullet screen comments from online TV series