Scholarlink

<p>The intangible cultural heritage (ICH) in China is currently being spread to the world in large numbers hence, the need to preach cultural sensitivity and preservation. A few data points and platform-specificity, as well as approaches mostly descriptive in nature or prescriptive, though, limit the current research, which, subsequently, discourages scalability and predictability. To overcome these weaknesses, this paper proposes the Greedy Politics Mutated Intelligent Convolutional Support Vector Machine (GP-ICSVM) model that incorporates CNN-based multimodal features extraction, GPO which is a feature selection strategy and kernelised SVM feature classification. To compile a multimodal dataset that contained 2500 records of (ICH in the form of images, videos, textual description, platform identifiers, heritage categories, and engagement metrics, social media sources and cultural repositories were used. Preprocessing of data included normalization, resizing, tokenization, uniform frame sampling (UFS), Term Frequency inverse Document Frequency (TF-IDF) vectorization and Z-Score standardization. CNNs captured hierarchical visual and semantic features, GPO prevented the redundancy of features, and the SVM estimated the effectiveness of international communication. TensorFlow/Keras was used to realize the GP-ICSVM framework to make it possible to reproduce multimodal processing and experiment. Empirical assessment shows better performance with an accuracy of 0.988 and engagement detection rate of 95.1%, which is higher than baseline CNN, RNN, Transformer and multimodal fusion models. The approach will allow taking ICH research to the next level of robust, predictive assessment and provide data-driven solutions to optimize the use of international digital promotion and the preservation of Chinese cultural heritage in the digital realm.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Analysis of the effectiveness and optimization of promotion strategies for the international communication of intangible cultural heritage based on convolutional neural network and SVM

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月29日

查看全文
  • Ya Li

摘要

The intangible cultural heritage (ICH) in China is currently being spread to the world in large numbers hence, the need to preach cultural sensitivity and preservation. A few data points and platform-specificity, as well as approaches mostly descriptive in nature or prescriptive, though, limit the current research, which, subsequently, discourages scalability and predictability. To overcome these weaknesses, this paper proposes the Greedy Politics Mutated Intelligent Convolutional Support Vector Machine (GP-ICSVM) model that incorporates CNN-based multimodal features extraction, GPO which is a feature selection strategy and kernelised SVM feature classification. To compile a multimodal dataset that contained 2500 records of (ICH in the form of images, videos, textual description, platform identifiers, heritage categories, and engagement metrics, social media sources and cultural repositories were used. Preprocessing of data included normalization, resizing, tokenization, uniform frame sampling (UFS), Term Frequency inverse Document Frequency (TF-IDF) vectorization and Z-Score standardization. CNNs captured hierarchical visual and semantic features, GPO prevented the redundancy of features, and the SVM estimated the effectiveness of international communication. TensorFlow/Keras was used to realize the GP-ICSVM framework to make it possible to reproduce multimodal processing and experiment. Empirical assessment shows better performance with an accuracy of 0.988 and engagement detection rate of 95.1%, which is higher than baseline CNN, RNN, Transformer and multimodal fusion models. The approach will allow taking ICH research to the next level of robust, predictive assessment and provide data-driven solutions to optimize the use of international digital promotion and the preservation of Chinese cultural heritage in the digital realm.