基于BP神经网络和模糊C均值聚类的多模态异构文化数据融合方法数据集

论文名称：Neural Network: An Improved FCM for Multimodal Cultural Data Analysis，针对文化、科技信息的多层面、多维度特性，提出了对文化大数据的特征提取、多模态数据拟合的技术方案，以神经网络方法为基础，对数据进行多模态的拟合，然后进行聚类融合。第一、对提取到的多模态文化数据的特征信息从地理、时间、艺术及主题人物维度进行分类汇总，形成多模态文化特征信息矩阵。
第二、从多个维度分别定义了多模态文化数据间关联度量化标准，并将数据集代入优化后BP神经网络进行训练，得到最终的关联度。
第三、使强关联性的文化数据聚集的模糊C-均值聚类策略，解决了数据无序的问题，为多模态文化数据融合问题提供了一种性能较优的融合策略。

Paper Title: Neural Network: An Improved FCM for Multimodal Cultural Data Analysis. Aiming at the multi-layered and multi-dimensional characteristics of cultural and scientific information, this paper presents a technical solution for feature extraction and multimodal data fitting of large-scale cultural data. Built upon neural network methodologies, the solution first performs multimodal fitting on the data, followed by cluster fusion. Specifically, 1) Classify and aggregate the feature information extracted from multimodal cultural data across four dimensions: geography, time, art, and thematic figures, to construct a multimodal cultural feature information matrix; 2) Define quantitative criteria for the correlation degree between multimodal cultural data from multiple dimensions respectively, and train the optimized Back Propagation (BP) neural network with the dataset to derive the final correlation degree; 3) Adopt a fuzzy C-means (FCM) clustering strategy that aggregates highly correlated cultural data, which resolves the issue of data disorder and provides a high-performance fusion strategy for multimodal cultural data fusion problems.

搜集汇总

背景与挑战

背景概述

该数据集针对文化大数据多模态异构特性，提出一种融合BP神经网络和模糊C-均值聚类的技术方案。它通过多维度特征提取和关联度量化，实现文化数据的拟合与聚类融合，以解决数据无序问题。数据集来源于国家重点研发计划项目，包含约73.93MB的文件，用于支持多模态文化数据分析。