Research On Emotion Recognition And Active Service Recommendation Method Based On EEG Signal

Emotions exist widely in all aspects of human life,and the research and application of emotion recognition also covers many fields.Compared with speech and face emotion recognition,EEG-based emotion recognition can directly reflect the activity of brain neurons and can provide more objective and accurate emotional information.However,EEG signals themselves are more complex,and in the process of emotion recognition,the extraction and utilization of signal features are often limited,resulting in low recognition accuracy.Therefore,how to effectively extract a variety of features in EEG signals and build a better emotion recognition model with deep learning methods has become the key to improving the accuracy of EEG signal emotion recognition.At the same time,under the circumstance that the recommendation algorithm is constantly developing but the active service recommendation based on emotion has not been fully studied,the main research work of this paper is as follows:(1)Aiming at the problem of how to effectively extract multiple features in EEG signals,this paper proposes a multi-feature EEG signal emotion recognition model,which mainly decomposes empirical modes of preprocessed EEG signals to obtain multiple connotative modal components(IMF),and then extracts the extracted IMFs components by Hilbert transform and cospatial mode algorithm respectively,synthesizes the obtained features and performs sample entropy conversion,and finally uses convolutional neural networks(CNNs)for classification.(2)Aiming at the problem of how to use deep learning methods to build a better EEG emotion recognition model,this paper combines the convolutional attention module(CBAM),CNN and bidirectional long short-term memory neural network(Bi LSTM)to propose a convolutional recurrent neural network model(CRNN-CBAM)based on attention mechanism.The model extracts the spatial information in EEG signals through CNN,uses the attention mechanism to learn the connection between EEG signal channels and spatial information weights,then uses Bi LSTM to extract the temporal information in the signal,and finally inputs Softmax for classification.(3)Aiming at the lack of research on emotion-based active service recommendation methods in service recommendation,this paper proposes a user service recommendation model based on user sentiment and Apriori algorithm,It mainly recommends services to users according to different EEG emotion recognition classification results.In summary,this paper takes EEG signals as the research object,and proposes a multi-feature EEG signal emotion recognition model and a convolutional recurrent neural network model based on attention mechanism.At the same time,an emotion-based active service recommendation method based on EEG emotions is proposed.Experimental results show that the proposed methods have good performance in their respective fields.