Article Preview
TopIntroduction
Design formal language symbols are divided into different forms. First of all, the information in graphic design works generally contains the hierarchy of primary and secondary, first subject and comprehensive, and then local and detailed. No matter the whole or part of information, they need to be displayed through the extension of the medium of design formal language symbols. Generally, the whole information symbol is composed of various language symbols with different or the same part of information through layout and reorganization. For example, points, lines and planes are called basic language symbols, which can form overall information such as images, characters and colors (Wang et al. 2021; Ji et al. 2021). Secondly, the design of formal language symbols also has personal subjective color. It can not only promote creators to carry out rich emotional expression of the picture, but also build a bridge of communication and understanding between creators and viewers. As a basic element of graphic design, how to artistic creation of formal language symbols should be highly valued.
For graphic design, point, line and surface are the main framework and basis of design. The use of formal language symbols of overall design is based on basic language symbols of point, line and surface. The composition of formal language symbols of overall design is also composed of elements of basic design formal language symbols. It also requires the overall design of formal language symbols to carry out information fusion (Meng et al. 2023). Graphic design graphics, images, text, color is the most important, the most common overall design formal language symbol, the overall formal language symbol is an indispensable element in graphic design, and also has a very important aesthetic significance. In the process of design practice, through the artistic processing of graphics, text, color, etc., the points, lines and surfaces that were originally independent and scattered in the picture are integrated to fill the whole visual space, making the design work more rich and concise, making the scattered layout form a whole object image, and the information of the work becomes integrated from the scattered.
Sentiment Analysis, which analyzes people's emotions or opinions based on the text they generate, has long been one of the most active areas of research in natural language processing. Identifying the underlying emotions expressed in a text is crucial to understanding the full meaning of the text. With the rapid development of social media platforms such as Weibo, Zhihu and Toutiao, people are increasingly sharing their views and opinions online. Sentiment analysis has attracted a lot of attention (He et al. 2022), because the opinions or emotions detected in the text are of great help to product recommendation, public opinion analysis, market prediction and so on.
The goal of document-level emotion analysis is to judge the emotion expressed by the whole document, such as a film review or a comment on a certain hot news. As long as the text to be analyzed exceeds the scope of a sentence, it can be regarded as document-level emotion analysis. A prerequisite for document-level sentiment analysis is the assumption that the opinions expressed in the whole text are directed at a single entity and contain the views of only one opinion holder.
In the traditional task of artistic emotion analysis, most models regard emotion analysis as a classification problem consisting of feature extraction and classifier training (Zehra et al. 2021). Initially, machine-learning-based methods using supervised classification or regression were used to train text models from polarity markers (Castellano et al. 2021). However, the performance of these models is largely dependent on a large number of manually processed features, such as affective dictionaries and other features with specific meanings.
With the proposed deep learning method, the performance of sentiment analysis model has been further improved. The most widely used Neural Network models in the field of emotion analysis include Convolutional Neural Network, (CNN), Bidirectional Long Short-Term Memory (BiLSTM) network (Mei et al. 2021), etc.