Ontology Reconstruction And Mapping For Multi-source Heterogeneous Data In Science And Technology

In recent years,with the rapid development of society,the quantity and structure of scientific and technological data have become more and more complicated.In order to better organized various historical data,meet business requirements,and improve the efficiency of data query and analysis,this article relies on the projects "Hebei Province Technology innovation big data standardization processing and application research and development" and "Hebei province science and technology innovation big data public service platform",proposed a method based on multi-source heterogeneous data ontology reconstruction and mapping method in the field of science and technology.Aiming at the reconstruction of science and technology ontology,this paper proposes a multi-source heterogeneous data ontology reconstruction model based on hybrid neural network text classification.In order to break through the bastion of data,an ontology mapping based on similarity matching of automatic encoder is proposed.The main work of this paper is as follows:(1)Ontology reconstruction of multi-source heterogeneous data based on Hybrid neural network text classification.By analyzing the data characteristics of scientific and technological big data,in order to solve the phenomenon of different data statistical standards and data clutter,the indicators have been combined and standardized,and detailed steps for standardization have been given;In order to solve the difficult problem of sharing data caused by the heterogeneity and dynamics of data,the difficulty of manual processing and the high error rate caused by the huge amount of data,this paper presents an ontology reconstruction method for multi-source heterogeneous data based on hybrid neural network text categorization.This paper proposes an ontology reconstruction method for multi-source heterogeneous data based on hybrid neural network text classification..In this paper,a multi-core convolutional neural network is designed to capture local features.At the same time,an improved bidirectional long-term and short-term memory network is added to compensate for the shortcomings of the convolutional neural network that cannot obtain context-related information.In order to improve the efficiency of the model,an attention mechanism is added.The preprocessed text data is trained on the word vector through word2 vec,the sentence matrix is obtained through the embedding layer,the sentence matrix is input into the model,and the deep learning model is trained,thereby realizing the ontology reconstruction.(2)Ontology mapping based on similarity matching of automatic encoderFirstly,the ontology mapping method based on traditional node matching is given.In order to reduce human participation and improve the accuracy of ontology mapping,an ontology mapping method based on deep learning similarity matching is designed.In order to avoid the disaster of dimension,dimension reduction is carried out according to the analysis of dimension weight.Then,dimension feature matrix is input into deep learning model.Considering the characteristics of big data in this paper,and in order to further improve the efficiency of the model and enhance the anti-interference ability of the model,the improved automatic encoder is integrated,and the ontology mapping method based on the improved deep learning similarity matching is proposed.By using the mechanism data set and the talent data set in the big data of science and technology,this paper compares the ontology mapping method based on the similarity matching of traditional nodes,the ontology mapping method based on the similarity matching of deep learning and the ontology mapping model based on the similarity matching of automatic encoder.The experimental results show that compared with the traditional method and the depth learning method,this paper proposes the following three methods The proposed ontology mapping method based on automatic encoder similarity matching has better performance.