Design Of Intelligent Question And Answer System Based On Deep Learning

At present,intelligent question answering system is a research hotspot in the field of natural language processing.It allows users to ask questions in natural language and can quickly and accurately return answers to users,eliminating the need for users to find out the answers themselves.In recent years,with the rapid development of deep learning technology,more and more researchers try to use short-text semantic modeling using deep neural networks,a computational model that is abstracted by simulating human cognitive processes.In this technical background,a set of intelligent question answering system is studied and designed based on deep learning in this paper.In this paper,the deep learning technology is used to design the intelligent question answering system.The main research work is divided into the following aspects: Firstly,in order to improve the retrieval performance of the system,the question classification model is constructed based on bi-directional long short-term memory network and attention mechanism in this paper,the model can effectively extract sentence semantic information and text features,and therefore has the advantage of sequence semantic modeling.Experimental results show that this model can more accurately classify the sentences than the traditional neural network model.Secondly,a sentences similarity measure model is built based on deep convolutional neural network,in which the combination of convolution and the maximum pooling operation has the function of local semantic extraction,and the deep network model extracts the key of more representative sentence semantics information.The comparative experiment shows that the model is effective in the tasks of sentence semantic feature extraction and similarity matching.Finally,based on the seq2 seq model,the traditional unidirectional recurrent neural network is improved by using bi-directional long short-term memory network,and introduces the attention mechanism again so that the model can seize the core semantic of the question to give more relevance Strong answer.Comparative experiments show that the dialog generation model can output higher quality answers on the decoding side.In the end,the modules are integrated and a set of intelligent question-answering system for the fund is designed.The mobile client of this system chooses the Android smart phone with higher penetration rate as the carrier of the application,and proves the system's effectiveness by testing the system performance.