Research On Prediction Of Consumption Intention Based On Concept Graph And Emotional Knowledge

Social media platforms are an important way to obtain human opinions,attitudes,intentions,and subjective inner world,and the related technologies such as natural language processing and sentiment analysis provide effective tools.This article predicts the consumer's consumption intention on the large-scale text data of Weibo,and finds that the user's consumer behavior will be affected by their emotion and sentiment.In the consumption intention recognition task,the paper focus on the impact of emotional information on the consumer intentions.The research content of this paper involves text emotion classification,intention domain classification and consumption intention recognition,which all provide effective technical support for the applications such as public opinion mining,dialogue system construction and consumption prediction.Therefore the research of this paper has important theoretical value and significance.The main research work of this paper is as follows:1.Text emotion classification based on emotional Word Vector.Traditional Word Vector based on the distributed hypothesis does not carry the emotional information of words,but the emotional attributes of words are a very important type of semantic knowledge,that is why the traditional Word Vector will directly affect the performance of emotion analysis.Therefore,this paper proposes a text emotion classification method based on Emotion Word Vector.Firstly,the Word Vector integrated with emotion information is generated.Then the emotion Word Vector is used to initialize the emotion classification model.Experiments show that the Emotion Word Vector can effectively improve the performance of the emotion classification task.2.Intention domain classification method based on concept graph and language model.The intention domain classification is one of the fundamental issues in the dialogue system.In view of the diversity and variations of users inputs in the dialogue system and the difficulty of obtaining large-scale corpus for intention domain classification tasks,this paper proposes an intention domain classification method based on concept graph and language model.This method explicitly applies the upper-level concept in the knowledge graph into the data processing and abstracts the users' intention to the concept level.Meanwhile,to reduce the ambiguity between the concepts,the language model is further used to rank all the concepts to choose the concept of the most suitable context.Because of the addition of extra information and knowledge,i.e.concept graph,the classification model can still maintain a high generalization rate on small-scale training datasets.And this method has less dependence on the domain and has good transplantability.Experiments show that the macro F1 value of the model on the SMP-ECDT dataset can reach94.21.3.Consumption intention recognition based on the concept graph and emotional knowledge.Because the users' consumption intention and behavior will be affected by their emotions,this paper researches the influence of emotion information on the task of consumer intention recognition and proposes a consumer intention classification model based on the concept graph and emotional knowledge.The method first uses the concept graph to process the input data and then the advanced features extracted by the emotion classification model is integrated into the consumer intention classification model.The experimental results show that the added emotional knowledge can help to improve the accuracy of consumer intention recognition,and when a domain contains more emotional expressions,emotional knowledge helps more.