A Fine-grained Emotional Analysis Of The Integration Of Ontology And Deep Learning

Fine-grained emotion analysis analyzes the author's emotion tendency from the perspective of the evaluation object and its attributes through the text.Its main tasks include the identification of the evaluation object and its attributes(subject recognition),emotion polarity recognition and the integration of subject and emotion.This paper proposes a fine-grained emotion analysis model that integrates ontology and deep learning.It combines ontology,emotion dictionary,deep learning and rules to identify fine-grained emotion of comment text.The main research contents include subject recognition,emotion polarity recognition and subject emotion fusion.Aiming at topic recognition,a method of combining domain ontology and neural network is proposed to identify the topic in text.Firstly,the stop words are removed by text preprocessing,and the result of text segmentation is obtained.Then,the explicit topic in text is recognized by the result of segmentation and domain ontology to ensure the accuracy of the explicit topic recognition,and the implicit topic in text is recognized by the result of segmentation and deep learning to ensure the availability of implicit theme,through the fusion method to identify the explicit theme and implicit theme in the text.Aiming at emotion polarity recognition,this paper proposes an emotion recognition method that integrates emotion dictionary into deep learning,uses emotion dictionary to identify emotion words,non-emotion words,degree words and negative words in text,designs word embedding model based on word type and part of speech,which makes neural network pay more attention to emotion words,degree words and negative words,and improves the effect of emotion recognition.At the same time,the attention mechanism considering the word location information is introduced to solve the problem of long-term dependence.Meanwhile,the semantic information around the target word is mainly considered based on the location information to improve the effect of emotion recognition.For theme and emotion recognition,an emotion analysis method is proposed,which integrates theme recognition and emotion recognition.In the process of theme recognition,it is interpretable.Explicit theme recognition can directly obtain the information of theme words in the original text.Implicit theme can obtain implicit theme vector to express implicit theme information through neural network.In the process of emotion recognition,explicit theme and implicit theme are integrated This paper focuses on the emotional words around these specific theme words,that is,adding the category and location information of theme words in the neural network to realize the emotional recognition of specific theme and further improve the effect of finegrained emotional recognition.Finally,through the experimental comparison,the effectiveness of the proposed fine-grained sentiment analysis model based on the integration of ontology and deep learning is proved.The experimental results show that the proposed fine-grained sentiment analysis method has certain advantages in accuracy,recall rate and F value compared with other methods.