A Research On Multi-source And Multi-granularity Education Examination Information Fusion Based On Deep Learning

Education information reform is related to the development of national education.With the continuous upgrading of education information deployment and the continuous improvement of education examination information construction,a large number of different types of education examination information systems have appeared in various provinces and cities across the country,such as adult college entrance examination information management system and student examination information management system.To actively coordinate the comprehensive reform of urban and rural education in Chongqing and adapt to its actual situation and needs,it is urgent to strengthen the research and construction of education examination information system,promote the vigorous and healthy development of education in Chongqing by integrating the existing heterogeneous data resources,and establish and improve the guarantee system of education examination information.But there are many in Chongqing through education examination information system at present,but the lack of an overall plan as a whole,the education administrative departments at all levels,miscommunication between education examination organizations,most of the existing education examination information system is set up by the education departments and institutions according to their needs,especially in the testing business system in Chongqing and Chongqing commission business system data docking is difficult,so it is difficult to carry out the cross-functional business process as well as the city's education examination information system analysis and comprehensive utilization of data.Therefore,it is of great practical significance to research information fusion of multi-source and multi-granularity educational examinations based on deep learning.Based on education examination information into the background,withoutchanging or as little as possible to change the original system and its business based on fully using the education information platform-independent distributed heterogeneous data,implement all kinds of education in Chongqing sharing of data exchange between the test information system,the granularity more source information fusion were studied,based on the deep learning of multi-source information fusion multi-granularity education examination.Firstly,the multi-agent multi-source heterogeneous data acquisition is studied,and its function and process are designed.Next,the multi-source information association based on CNN is studied,the information association algorithm is designed to customize the data set,and the architecture of the CNN model is designed.Then,the multi-granularity feature fusion based on CNN and LSTM was studied,and the data set was customized by the granularity computing algorithm,and the multi-granularity feature fusion was carried out by the combination of CNN and LSTM.After that,based on the above research on the multi-agent acquisition system and data fusion,this thesis constructs the multi-source education examination system and the multi-source information fusion database of education examination.At last,it summarizes the research work of this thesis,puts forward some shortcomings of the research content,and looks forward to the future research direction.In this paper,we also has carried on the simulation,analysis and verify the effectiveness of multi-source education examination system,the experimental results show that this design is based on CNN's multi-source information correlation and multiple granularity characteristics fusion based on CNN and LSTM has higher accuracy,the use of the system can be integrated with full now scattered in various departments of education examination data resources,improve the utilization rate of data resources and promote the development of information-based education examination.