RESEARCH-ARTICLE
Audience Feedback Assessment System for International Communication of Dongguan Intangible Cultural Heritage Using Multimodal Sentiment Analysis CNN-BiLSTM Model
This research presents a novel multimodal cross-cultural sentiment analysis framework for assessing audience feedback on Dongguan Intangible Cultural Heritage (ICH) video promotion on social media platforms. The proposed CNN-BiLSTM fusion architecture, integrated with a Cross-cultural Sentiment Calibration (CCSC) layer, combines textual comments, visual features, and cultural embeddings to achieve equitable and accurate sentiment classification across diverse cultural groups. It significantly improves F1-score to 87.3% and reduces performance disparity across language groups, leading to a Cultural Fairness Index of 0.94. Case studies reveal distinct cross-cultural reception patterns for different ICH types, providing actionable insights for promotion strategies.
Impact Metrics
Our advanced AI system delivers measurable improvements in sentiment analysis accuracy and fairness for cultural communication.
0
F1-score Achieved
0
Improvement vs. Text-Only
0
Cultural Fairness Index
0
Reduction in Inequality
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Multimodal Fusion
Cultural Calibration
Case Studies & Insights
Multimodal Fusion Boosts Accuracy
Integrating visual and textual features is critical for comprehensive ICH video understanding. The CNN-BiLSTM architecture significantly outperforms unimodal approaches.
0
Percentage Point Improvement over Text-Only Baselines
CNN-BiLSTM Multimodal Fusion Model Flow
Our proposed architecture integrates visual and textual modalities through a dual-branch fusion framework. The model consists of three primary components: a visual feature extraction branch, a textual feature extraction branch, and a multimodal fusion module.
Enterprise Process Flow
Video Keyframes
→
ResNet-50
→
Temporal Pooling
→
Visual Features
→
Comment Text
→
mBERT
→
BiLSTM
→
Textual Features
→
Bilinear Pooling
→
Multimodal Fusion
CCSC Layer Ensures Fairness
The Cross-cultural Sentiment Calibration Layer addresses cultural bias, dynamically adjusting feature weights based on cultural context and ensuring equitable performance across diverse language groups.
0
Achieved Cultural Fairness Index
F1-scores Across Cultural Groups (With vs. Without CCSC)
The CCSC layer significantly reduces performance disparity across different cultural groups, leading to substantial improvements for non-English speaking audiences.
| Language Group | Without CCSC | With CCSC | Improvement |
|---|---|---|---|
| English | 85.3% | 88.1% | +2.8% |
| French | 78.6% | 86.9% | +8.3% |
| Japanese | 74.2% | 86.5% | +12.3% |
| Other Languages | 72.1% | 85.8% | +13.7% |
| Standard Deviation | 5.8% | 1.1% | -81.0% |
| Cultural Fairness Index | 0.68 | 0.94 | +38.2% |
Dragon Boat Racing: Universal Appeal
Dragon Boat Racing achieved the highest positive sentiment (92.3%) and acceptance (89.1%), demonstrating that its competitive nature and visual spectacle transcend cultural boundaries.
Dragon Boat Racing: Universal Appeal
Challenge: Identify ICH with broad cross-cultural appeal.
Solution: Focus on visually dynamic and competitive aspects for international marketing.
Result: Highest positive sentiment and acceptance rate among all ICH categories studied, indicating strong universal appeal.
Cantonese Opera: Language & Cultural Barriers
Cantonese Opera showed polarized reception (68.2% positive, 43.7% acceptance), with language barriers and cultural context issues being primary obstacles, despite appreciation for visual elements.
Cantonese Opera: Language & Cultural Barriers
Challenge: Overcome significant language and cultural barriers for performance-based ICH.
Solution: Provide real-time translated subtitles, educational content, and focus on visually stunning excerpts for initial exposure.
Result: Revealed visual appreciation but auditory and contextual struggles for non-Cantonese speakers, necessitating enhanced accessibility strategies.
Advanced ROI Calculator
Estimate the potential efficiency gains and cost savings for your enterprise by implementing an AI-powered sentiment analysis system.
Annual Savings
$0
Hours Reclaimed Annually
0
Implementation Timeline
A structured approach to integrating AI into your enterprise, ensuring a smooth transition and rapid value delivery.
Phase 1: Data Integration & Model Setup
Collect and preprocess social media data, set up the CNN-BiLSTM architecture, and integrate the CCSC layer. Est. Duration: 4-6 Weeks.
Phase 2: Adversarial Training & Calibration
Train the model with adversarial learning to mitigate cultural bias and fine-tune for optimal cross-cultural fairness. Est. Duration: 6-8 Weeks.
Phase 3: Deployment & Real-time Monitoring
Deploy the system for real-time audience feedback analysis and integrate with existing marketing platforms. Est. Duration: 3-5 Weeks.
Phase 4: Strategy Optimization & Iteration
Utilize insights from the system to refine content strategies and continuously improve international communication effectiveness. Est. Duration: Ongoing.
Unlock the Full Potential of Your Global Communication
Schedule a personalized strategy session to discuss how our AI-powered sentiment analysis can transform your international outreach and cultural preservation efforts.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat