Data Processing Method for Building Material Repository Based on Deep Learning and Model Training
In this paper,we explore an intelligent data processing method for construction material databases and unstructured data based on deep learning and Generative Pre-trained Transformers(GPT).The research aims to enhance the efficiency and accuracy of construction material data management,reduce the complexity of manual operations through automated processing,and save time and costs for enterprises.Various techniques were employed,including word embeddings,TF-IDF,cosine similarity,Named Entity Recognition(NER),and GPT-based fuzzy matching.Our experimental results demonstrate the outstanding performance of this method in material data cleansing,alias matching,information extraction,and form generation,exhibiting high accuracy and efficiency.This study presents an innovative data processing solution for the construction industry,with the potential to play a significant role in engineering project management.
construction enterprisesmaterial managementcost managementdata cleaning
万方数据
维普
