Automatic Recognition Of Classical Poetry Artistic Conception

Poetry is the jewel in the crown of China’s classical culture and has been praised and studied by countless people for thousands of years.However,due to the lack of technical methods,the study of classical poetry has always relied on the human labor of scholars and scholars.In recent years,with the rapid development of computer technology and the rapid improvement of hardware computing power,natural language processing technology has achieved remarkable results in practical operation.In this paper,natural language processing is applied to the text analysis of classical poetry,and a set of methods for automatic recognition of the artistic conception of classical poetry is proposed,which will contribute to the batch study of classical poetry,strengthen cultural self-confidence,and inherit and carry forward the excellent traditional Chinese culture.A large number of texts of ancient poetry and non poetry ancient Chinese are collected from the Internet asynchronously through customized writing crawler.The corpus is cleaned and put into storage.In order to be as close to the original Chinese as possible,traditional Chinese is used as the standard in the process of data collection and research.Word embedding is used to train word vector and complete the preparation of corpus.This paper studies the application of different machine learning algorithms in text classification,combines them with different document vectorization methods,compares their performance in poetry topic classification,and summarizes that the classical machine learning framework has better accuracy.By comparing the effect of word based vector and word based vector,it is concluded that word based vector has a high accuracy.The deep learning method is further introduced into the research to analyze the advantages and disadvantages of all kinds of neural networks.The research community has a good effect on the neural network architecture in the practice of Natural Language Processing,such as Text CNN and Bi LSTM,etc.,and the mature Natural Language Processing pre training model,such as BERT,is introduced for topic classification.An emotion dictionary matching method based on word vector is constructed for emotion analysis.This paper studies the different expressions of ancient Chinese and modern Chinese in natural language processing,analyzes the difficulties in word segmentation and vector construction of ancient poetry,solves the difficulties in word segmentation through single word standard,and solves the problems in lack of reliable methods for directly constructing document vector through studying the mapping from word vector to document vector.The automatic recognition algorithm is applied to a large number of unmarked poems.Based on the obtained data,the system of poetry mood analysis and similar poetry recommendation is developed.After the user inputs a poem,the result of poetry artistic conception will be inferred and similar poetry will be recommended.The poetry in the database can also be classified and queried through theme and emotion screening.