Research On Intelligent Question Answering System Based On Deep Learning

Since the beginning of the 21 st century,science and technology have achieved rapid development.Computer technology has profoundly affected people,and people's daily lives have become increasingly dependent on computer technology.Especially in the past two or three years,with the rapid development of computer software and hardware technology,the computing speed of computers has become faster and faster,and the storage capacity has also become larger.Artificial intelligence technology has also obtained new development opportunities,and it has obtained new development opportunities in many fields.It has a wide range of applications.Natural language processing based on artificial intelligence refers to the use of computers to recognize,process,understand and correctly respond to human language.Intelligent question answering system is a research hotspot and difficult point in the field of natural language processing,and it has broad application prospects and practical significance.This topic designs and implements a Chinese natural language intelligent question answering system based on deep neural networks,and conducts in-depth analysis and research on some key issues of Chinese word segmentation and semantic understanding of the intelligent question answering system,which greatly improves the question and answer model.Performance and accuracy.Chinese is different from English.English words are originally separated by spaces.Chinese words must be segmented with the help of context.Aiming at the problem of insufficient understanding of context words in Chinese word segmentation,this topic improves the Chinese word segmentation method in deep neural network.Specifically,the Chinese word segmentation model based on two-way LSTM is optimized,and different weights are set for the forward LSTM layer and the backward LSTM layer respectively,so as to enhance the capability of the network and better improve the accuracy of Chinese word segmentation.In addition,this paper proposes a machine reading comprehension model with a multi-level attention mechanism,which can simulate the understanding process of humans from shallow to deep,from rough to detailed,from summary to detail when reading an article,and combines different types of attention mechanisms.It is applied to multiple network layers to capture the relationship between the question and the article at different levels of granularity,gradually focus on the boundary of the best answer,and finally predict the correct answer by deliberating the details.And through multiple sets of experiments on different data sets,the effectiveness of the model is verified.Finally,this paper implements a Chinese intelligent question answering prototype system based on deep learning.The entire prototype system uses a classic three-tier architecture design,from top to bottom are the interface layer,logic layer and training layer.The functional architecture is divided into user interaction module,task scheduling module,data preprocessing module,Chinese word segmentation module,answer prediction module and model training module.The user asks a question to the system,and the system returns the calculated answer to the user after calculation.The entire question and answer process can be basically real-time.