Research On Sentence Similarity Calculation Method Based On Deep Learning

With the popularity of computers,people are accustomed to using the Internet to obtain the information they need,but because digital information is increasing exponentially year by year,when people use search engines,search engines can only return links to related information or contain answers.Large texts,users want to get useful information and secondary screening,search engines can not meet the needs of most people.The answer to the user's answer to the question and answer system is not a whole bunch of related documents or related links,but a more precise answer,which is more in line with the user's needs.The key technology of the question and answer system is the question similarity processing.The traditional similarity research method can not effectively capture the semantic information of the sentence,and the accuracy of the matching is not high.The convolutional neural network in deep learning can effectively extract the sentence features,and the cyclic neural network can capture the context information of the sentence.Both methods can capture the semantic information of the sentence well,and use the sentence similarity accuracy of deep learning.higher.This paper compares the advantages and disadvantages of convolutional neural network(CNN)and cyclic neural network(RNN)in text processing in deep learning.Combining the knowledge of deep learning,the deep matching semantic model is studied,and the semantic matching model based on convolutional neural network is studied.The k-max mean sampling is used in the pooling layer,and the sentence based on k-max mean sampling technique is proposed.The similarity algorithm,compared with the original model,shows that the method used in this paper is more accurate.In the course of experimental research,it was found that this method lackedthe ability to capture contextual semantic information.Aiming at the problem that convolutional neural network can't capture the semantic information of text context,a method for calculating the similarity of sentences based on LSTM and convolutional neural network is proposed.Firstly,the semantic information of the context is captured by the cyclic neural network,and then the convolutional neural network is used.The features are extracted,and finally the cosine similarity is used for sentence similarity processing.This method combines the advantages of the first two methods by using the root mean square error as the evaluation criteria for the experiment.The experimental results show that the proposed method is more accurate.