The Application Of Convolutional CIFG Model In Sentiment Analysis Of Film Reviews

With the rapidly development of internet technology,online commentary plays an important role at both the national level and in people's daily lives.The analysis of text on web is a realistic task.Our Chinese characters are full of rich semantic information.Sometimes,the same words have different meanings in different contexts,and the studies of Chinese also requires word segmentation,translate the unsimplified Chinese character into simplified Chinese character and so on.The Chinese text analysis is more challenging than English text analysis.Text sentiment analysis is a non-negligible part of text analysis.Usually we have to understand the emotion of a sentence,a comment,or a paragraph to understand the true meaning of the sentence more accurately.Artificial intelligence is gradually changing people's lives.Deep learning plays a role in promoting artificial intelligence.The face recognition,voice dialogue and text analysis techniques developed on deep learning are being applied to the App of our mobile phone.The analysis of Text sentiment analysis is an important application of text analysis technology.Among many neural networks,convolutional neural networks have local perceptual ability to extract local information of data,and recurrent neural network are more suitable for text data studying.The language is serialized data and the language context has semantic meaning.Therefore,in this paper I combines these two neural network and proposes a convolution neural network with coupled input and forget gate network(CNN-CIFG)to solve the emotional analysis problem in film review text.In this paper,firstly I crawls the film review text information from Douban online film and television review through web crawler technology.Because of film review data has the character of random and diversity,it is necessary to preprocess the film review text such as data cleaning,text unsimplified and simplified conversion,text segmentation and do the emotion label.The textwhich has been preprocessed is vectorized,and the film review text is converted into a form that the computer can recognize,and then the vectorized data is transmitted to the neural network model.After continuous iterative optimization,a model that can be used for filming emotional classification is obtained.Through experimental comparison and analysis,the CNN-CIFG model is constructed in this paper and it can learn the emotional information in text better than others,the classification performance both positive and negative texts is improved in multiple evaluation indicators.