Application Of Improved Convolutional Neural Network Models In Text Classification

In recent years,with the rapid development of computer hardware and the Internet society,the volume of information people receive and create has grown exponentially.It has greatly contributed to the development of deep learning technology and brought about a new wave of artificial intelligence.In this wave of development,natural language processing has become a key research direction in the field of artificial intelligence,aiming to study the relationship between humans and machines using Theories and methods for effective communication in natural language.Communicating with computers in natural language has very important practical applications and revolutionary theoretical implications.Modern deep learning algorithms commonly used for text classification include the network model of recurrent neural networks(RNNs)and their variants,convolutional Network models of neural networks(CNNs)and their variants offer better performance and results than traditional machine learning methods,but Existing models of convolutional neural networks do not take advantage of the vast amount of knowledge that humans have accumulated over the years,and at the same time,for words within the same sentence and The mining of intrinsic connections and dependencies between words is also inadequate.In this paper,we address these deficiencies of convolutional neural networks on text classification tasks and propose two improved posterior approaches for different tasks the new model.A convolutional neural network model incorporating a sentiment dictionary is first studied and designed and used to handle the task of sentiment classification of text.Firstly,the concepts and knowledge related to classical text classification methods are introduced in detail,followed by the concepts and knowledge related to the processing of Chinese text data with English Differences and difficulties in textual data.To address the inadequate use of prior human knowledge in emotion classification tasks,the emotion lexicon is combined with a convolutional neural network model.A new model for classifying sentiment in Chinese text is proposed,and its basic ideas are elaborated in detail using a large number of real datasets.The validity and superiority of the model is validated.Secondly,a convolutional neural network model incorporating the attention mechanism is investigated and improved and used to handle the text multiclassification task.Firstly,the relevant concepts and principles of the attentional mechanism are introduced in detail,and secondly,the existing deep learning models are analyzed for their performance in processing Chinese text multi-categorization tasks.deficiencies when classifying problems.This paper uses an attentional mechanism to improve and optimize the convolutional neural network model to better capture the meaning and features represented by sentences the classification effect was optimized and the effectiveness and superiority of the algorithm was verified experimentally.