Machine learning-based multidimensional sentiment visualisation and analysis of digital media

Digital media contains a huge amount of emotional information that needs to be mined. To solve the problem that existing models ignore the features of multi-dimensional emotional words, firstly, multi-dimensional emotional words are expanded based on improved Word2vec, and then the digital media comments are input into the pre-trained model to generate a multi-dimensional text emotional word vector. The modified term frequency-inverse document frequency (TF-IDF) method is used to obtain the representation of multidimensional emotion subject words. Then the global features are obtained by using the hybrid model of convolutional neural network (CNN) and gated recurrent unit (GRU). Multi-dimensional attention mechanism is used to interact global features and multi-dimensional emotion features, and multi-dimensional emotion classification results are output by full connection layer. The results show that the Marco-F1 of the proposed model is 91.17%, which can accurately classify the emotions of digital media.