National Image Construction Based on Multimodal Translation in the Context of Big Data

Abstract

This study focuses on the public’s perception of a specific country’s image, and adopts comment texts from multimodal translated materials as the object of analysis. A country image text perception model was established by combining the TextRank algorithm and the LDA topic model. The model extracts and categorizes themes from the collected comment data, and discusses in depth the different themes of country image and their degree of attention. It is found that the most popular theme is “economic dimension”, with a heat degree of 42.59%, followed by “regional development” and “political dimension”. This reflects the high level of public interest in these areas. In addition, the study quantified the public’s emotional disposition towards the country’s image and found that most people have a positive attitude towards the country, with a high percentage of positive sentiment of 84.89%, but most of them are mildly positive. This suggests that although the public’s overall perception tends to be positive, there is room to improve the country’s image and enhance the public’s positive perception.