User Travel Domain Intention Prediction Of Question Answering System

In recent years,with the continuous development of deep learning technologies,smart products such as question answering systems and chatbots have also appeared one after another.Question answering systems have become a hot research area in the field of natural language processing(NLP)because they can quickly and accurately feedback the collection of user questions and answers.Question answering systems are divided into many types,and different types of question answering systems have different ways of processing data.However,according to the flow of data,the question answering system can be roughly divided into three parts,namely semantic understanding,information retrieval and answer generation.In this paper,the semantic understanding part of the question answering system is placed in a specific travel area to identify the user's intention in the travel area.In view of the current complex and time-consuming process of intent recognition,this paper introduces the shallow neural network FastText model to study the field of short text travel for users.One of the main innovations of the FastText shallow neural network is the introduction of N-gram features to solve the problem of loss of word order,as the input word order is processed by N-gram,a large number of meaningless redundant words will be generated.TF-IDF is used to select the weight of the word sequence to generate a reserved dictionary.In addition,for short text content and sparse features,this article introduces the LDA topic model for topic feature word selection,and expands the feature dictionary to improve the accuracy of intent recognition.In addition,since the computer cannot directly understand the language,in order for the computer to process the text,it needs to be converted into a data structure that the computer can recognize.After studying the structure and internal form of Chinese characters,this paper proposes a Chinese word vector model GCWE that combines radicals and Chinese characters.Based on the CBOW model,the CBOW model is combined with the radical information of Chinese characters to improve the CBOW model Make it more suitable for the field of Chinese text.Finally,this paper applies a short text travel domain intent recognition model to a question answering system,and identifies the travel intention of the user by identifying the travel domain text.After system testing,the average recognition accuracy of travel intention module in the intelligent question answering system module is over 90%,and contributed to the information retrieval and answer generation of the intelligent question answering system in the later stage.