Research On Military Machine Reading Comprehension Technology Based On Deep Neural Network

Machine reading comprehension is a very popular research direction in the field of artificial intelligence.Its purpose is to find the answers to user’s questions accurately from a given article through a series of operations,such as semantic analysis,retrieval and evaluation of article content.Machine reading comprehension is the biggest challenge in natural language processing technology after speech judgment and semantic understanding.Deep learning is an important branch of machine learning.Its motivation is to build and simulate the neural network of human brain for analysis and learning.It imitates the mechanism of human brain to interpret data,such as images,sounds and text.Military personnel sometimes come into contact with a large amount of information,due to time constraints can not read one by one,but at this time there are more questions to be answered.If professional tools can be used to assist staff to understand information and give answers to simple questions,useful and accurate information and knowledge can be obtained in a relatively short time,which will undoubtedly reduce the workload of staff and provide strong support for extracting effective information from the information.The purpose of this thesis is to study military machine reading comprehension technology based on deep neural network.Firstly,the data set is analyzed,and the data set is pretreated by text truncation and answer tagging according to the analysis results.Then,we use context based dynamic word vector representation to represent text and problem.Then the accuracy of the single model is improved by training the machine reading comprehension model based on bi-directional attention mechanism,the machine reading comprehension model based on bi-directional long-term and short-term memory network and the machine reading comprehension model based on self-attention mechanism and convolution neural network.Then use ensemble learning,give full play to the advantages of each model,improve the accuracy of the model through the way of bagging,and finally standardize the format of the answer,further improve the accuracy.