| Now,4G networks are becoming more and more popular.WIFI can also be seen everywhere.As a result,it is becoming easier and easier for people to get information.When we suffer from problems,we usually rely on search engines to find solutions.Ordinary retrieval systems often provide us with a lot of information.There is very little valuable information.It is very difficult for us to extract information.Through this question and answer system,we can solve this problem quickly and effectively.For those who have just learned to drive.The most important thing is how to behave correctly.For safety reasons,we should drive according to the instructions.As the pace of life speeds up,we rarely have large chunks of time to learn fully.We hardly have large chunks of time to study adequately.So we can use the question and answer system.It is ready to provide us with standard answers to our problems.This paper designs an intelligent question answering system in a limited field.In this paper,the questions and answer in the instructions are made into data samples.We compared the user’s question to this sample.Calculate the similarity between them.Let’s put them in a similar order.We find the problem with the highest degree of similarity.Let’s find out the answer to this question.Finally,it feeds the answer back to the user.In this paper,the question answering system mainly studies the matching between the user’s questions and the sample candidate questions.There are several difficulties in completing this work.The first is how to make better use of samples.The second is how to segment keywords.The third is how to adjust the imbalance between positive and negative samples.We have completed the above questions..In this paper,cyclic neural network,TFIDF algorithm and bidirectional long-term and short-term memory network are used to model.A question-and-answer sample was constructed manually.The open source package of "jieba" is also used to segment the question and answer samples on Python.The semantic similarity model based on TFIDF and the semantic similarity model based on Siamese-BiLSTM are established respectively.The results show that the performance of the model based on Siamese-BiLSTM is better than that based on TFIDF.The accuracy of the semantic similarity model based on Siamsese-BiLSTM is 99.52% on the test set. |
