Research On Classification Model Of Civil Aviation Uncivilized Passengers Using Mixed Text Data

Aviation safety is an eternal topic of civil aviation.The classification of uncivilized passengers in civil aviation is an important part of aviation safety.Countries around the world have gradually established a civil aviation uncivilized passenger classification system to identify in uncivilized passengers.Different security measures are being taken in passengers of different dangerous levels.Passenger data,online records,public security records,travel records,telecommunication records,bank records,public opinion records etc.are used in the classification process of uncivilized passengers.These data are collectively referred to as passenger records.At present,the main problem encountered in the classification of uncivilized passengers in civil aviation is the feature extraction of text data in civil aviation passenger records.Especially the mixed text in text data,because the existing natural language processing method is mainly for monolingual text,there are many problems when dealing with mixed text.Therefore,it is extremely urgent to study the feature extraction method of mixed text to improve the classification accuracy rate of uncivilized passengers in civil aviation.In view of the above questions,this paper first proposes a hybrid text feature extraction and classification method based on deep learning model.This method combines bilingual vector and two-channel deep learning network.We uses convolutional neural network and attention mechanism to extract local key features of Chinese vector.The Bi-directional Long Short-Term Memory and attention mechanism are used to extract the global key features related to the English vector context.Finally,the text features extracted by the two complementary models are combined and input into the classifier for classification,which effectively solves the problem of difficult feature extraction of mixed text.On this basis,the text data extraction method for civil aviation records is further studied,and a classification model based on combined features is proposed.This model is a three-channel deep learning network.The two-channel deep learning network is used to extract the text data features in the passenger record,and a channel is added to extract the category numerical data features.Finally,the features extracted by the three channels are combined and classified.The experimental verification shows that the uncivilized passenger recognition rate of the model reaches 90.9%,and the unclassified passenger classification accuracy rate reaches 92.8%,which effectively improves the classification accuracy of uncivilized passengers in civil aviation.