Research On Knowledge Service And Interactive Visualization Component Of Cross-Media Big Data Of Science And Technology

In the era of big data,scientific and technological resources present trends such as large-scale data,multi-modality,rapid update,high timeliness and low-value density,which pose severe challenges to the effective use of scientific and technological resources.Based on knowledge and technologies such as attention mechanism,adversarial generative learning,recurrent neural networks,and microservices,this thesis uses deep learning methods to perform entity recognition and relationship extraction on scientific and technological big data,introduces a multi-attention mechanism to improve the accuracy of the algorithm;uses adversarial generative learning Realize cross-media big data of science and technology retrieval with semantic similarity;use the cyclic neural network method to take the entities and relationships in the knowledge graph as input to deduce new relationships and further expand the scale of the knowledge graph.At the same time,the development of knowledge service components such as entity-relationship extraction,cross-media retrieval,and interactive visualization is completed through microservice technology,and the open collaboration of knowledge service components is realized.The main tasks completed in this thesis are as follows:(1)A method for entity recognition and association discovery of cross-media technology big data is proposed.Aiming at the entity recognition of scientific and technological big data,a named entity recognition algorithm combining word segmentation part of speech attention mechanism is proposed.By combining word segmentation part of speech,two-way long and short-term loop network and attention mechanism supervised method is used to learn and Through training,the entity recognition of scientific and technological big data is realized;for the entity-relationship discovery of scientific and technological big data,by combining the sentence-level attention mechanism and self-attention mechanism of scientific and technological text data,an entity-relationship extraction algorithm based on the multi-attention mechanism is proposed.Realize the entity-relationship discovery of scientific and technological big data.(2)The open collaboration of knowledge service components of cross-media technology big data is proposed.In order to make the knowledge service component have good scalability and openness,an open collaboration mechanism based on the microservices of the knowledge service component is proposed,and the specific knowledge service component developed in this thesis is defined at the same time.Use the microservice architecture to complete the development of knowledge service components,realize the scalability,openness,and distributed operation of knowledge service components,and have the advantages of fast business response,high code reuse rate,high reliability,and low development cost.Aiming at the cross-media big data of science and technology retrieval knowledge service component,a cross-media big data of science and technology retrieval method based on confrontation generation learning and semantic similarity is proposed,which retrieves cross-media technology resources and sorts them according to semantic similarity,which solves the problem The heterogeneity of media data characteristics improves the accuracy of cross-media retrieval.(3)A method of dynamic deduction and interactive visualization of cross-media technology big data is proposed.Aiming at the problem of the lack of entity relationships in big data of science and technology,a dynamic deduction algorithm based on recurrent neural networks is proposed.Using the structure of the recurrent neural networks,the entities and relationship vectors in the knowledge graph are used as input,combined with the state of the previous moment.Calculate to obtain the hidden value at the current moment,and after the iterative operation,output a result vector that combines the entities and relationships in the knowledge graph,which solves the problem of the lack of entity relationships in the big data of science and technology,and expands the scale of the knowledge graph of big science and technology data.Aiming at the complexity of the internal relationship of cross-media big data of science and technology,using visualization techniques to help users obtain the hidden value behind the data,an interactive visualization scheme based on ECharts for cross-media big data of science and technology is proposed,which realizes the knowledge map of cross-media big data of science and technology.The visualization of,hot word cloud,etc.provides convenience for users to analyze big data in science and technology.(4)The system of cross-media technology big data knowledge service and interactive visualization component is realized.The system is composed of knowledge service components such as entity-relationship extraction,cross-media retrieval,and interactive visualization.The system has complete functions and a friendly user interface.