Clustering and Analysis of Dongguan Red Cultural Heritage Site Data Based on Support Vector Machine (SVM) and Locally Linear Embedding (LLE)

With the rapid development of data science and artificial intelligence, high-dimensional data classification has become a critical challenge in various fields, including cultural heritage analysis. Traditional classification methods struggle with high-dimensional feature spaces, leading to computational inefficiencies and suboptimal classification performance. In particular, the categorization of red cultural heritage sites in Dongguan involves complex multi-dimensional features such as geographical, textual, and visual attributes, necessitating an effective dimensionality reduction approach to improve classification accuracy while maintaining structural integrity. To address this issue, this study proposes a Locally Linear Embedding (LLE) and Support Vector Machine (SVM) hybrid framework for high-dimensional data classification. LLE is employed to reduce dimensionality while preserving local geometric structures, enhancing the separability of data in the feature space. Experimental results demonstrate that LLE+SVM achieves superior classification performance, with a maximum accuracy of 94.1%, significantly outperforming PCA+SVM and direct high-dimensional SVM classification. Additionally, computational time analysis confirms that LLE effectively reduces processing overhead while enhancing classification efficiency.