Semantic enhanced multichannel bidirectional graph convolutional network for aspect level sentiment analysis

The invention discloses a semantic enhancement multi-channel two-way graph convolutional network for aspect-level sentiment analysis, and relates to the technical field of semantic sentiment analysis, comprising: model training: learning and training data to enable a system to learn key features and rules from the data, and applying the key features and rules to new data; sentiment analysis: analyzing the probabilities of aspect words in various sentiment polarities. According to the semantic enhancement multi-channel two-way graph convolutional network for aspect-level sentiment analysis, BiLSTM is connected through random dense residual errors, deep-level semantic features can be extracted from shallow-level semantic features, semantic feature representation is enhanced, the problem of overfitting of stacked BiLSTM is well solved, various relation features among words are considered, and the semantic enhancement multi-channel two-way graph convolutional network for aspect-level sentiment analysis is more ac