基于深度学习的湖南地区清朝古桥可视化研究

Visualization of Qing Dynasty ancient bridges in Hunan area based on deep learning

近年来,随着人工智能相关技术的飞速发展,机器学习和深度学习等技术在计算机视觉和自然语言处理等领域都有了巨大的突破。利用人工智能相关技术从大量零散的古籍文本中挖掘有效信息,可以在保持人工成本的前提下,极大提高建筑类古籍的利用率,促进历史建筑的古籍文献基础研究。将采用基于卷积神经网络的Bert-BiLSTM-CRF模型图像分类方法和基于BiLSTM-CRF的命名实体识别方法实体方法对湖南清代地方志古籍进行古桥相关信息提取。建立数据库并将挖掘出桥梁相关的有效信息从定性和定量两个方面进行可视化研究。总结出桥名的命名方式、古桥建设情况,并结合ArcGIS进行空间分布特征分析,为历史建筑文献研究和古籍挖掘提供新思路。

In recent years,with the rapid development of artificial intelligence-related technologies,machine learning and deep learning technologies in computer vision and natural language processing have made great breakthroughs.Using artificial intelligence technology to mine effective information from a large number of scattered ancient books can greatly improve the utilization rate of architectural ancient books while keeping the labor cost,promote the study of the documentary basis of historical buildings.In this paper,we will use the convolutional neural network Bert-BiLSTM-CRF model image classification method and the BiLSTM-CRF-based entity recognition method to extract the bridge-related information from the local chronicles of the Qing Dynasty.The database is set up and the effective information about the bridge is mined from the qualitative and quantitative aspects.This paper summarizes the naming method of the bridge name and the construction of the ancient bridge,and analyzes the spatial distribution characteristics combined with ArcGIS.It provides a new way of thinking for the research of historical building documents and the excavation of ancient books.