A Semantic and Syntactic Enhanced Neuromorphic Computing System and Its Application in Consumer Sentiment Analysis

Consumer sentiment analysis can help users provide objective and clear recommendations through an amount of data, which is conducive to overcoming the ambiguous weakness in human subjective judgment. However, the precision, computational efficiency and robustness of existing consumer sentiment analysis methods still need to be improved. To address these issues, this paper proposes a semantic and syntactic enhanced neuromorphic computing system (SSE-NCS) for consumer sentiment analysis. Specifically, three main components are used to facilitate the circuit design of the proposed SSE-NCS. Firstly, the attention module consisting aspect-aware attention unit and self-attention unit is proposed to capture the information interaction. Secondly, the syntax-mask module is constructed to obtain the syntactic structure information. Thirdly, the graph convolutional module is designed to enhance the node representation and obtain the reliable output. Finally, the proposed SSE-NCS with hybrid training method is applied to consumer sentiment analysis, and the experiment results demonstrate that the proposed SSE-NCS has good performance in terms of classification accuracy, computational efficiency, and robustness.