Author(s)

Siyong Wu

Corresponding Author

Siyong Wu

Abstract

In the modern entertainment industry, the career of an acting star is closely related to his/her public image and reputation, and the consistency between the two plays a crucial role in the star's market performance and career development. However, current script selection and characterization rely mainly on the subjective judgments of directors, casting agents, and producers, and there is a lack of systematic tools to scientifically assess the fit between a star's image and a scripted character. This approach is not only susceptible to bias, but also suffers from decision-making risk and uncertainty. In order to solve this problem, this study proposes a BERT (Bidirectional Encoder Representations from Transformers) based model for assessing the consistency between scripts and stars' images, aiming to analyze the similarity between script texts and stars' established public images through natural language processing techniques. In this study, a large amount of script data and star image description data were collected, and the semantic features of the text were extracted by using the BERT model through preprocessing steps such as data cleaning, word splitting and context embedding. Then, the cosine similarity is used to calculate the similarity between the script and the star image, and the model is trained and consistency evaluated by supervised learning. The experimental results show that the model performs well in several indexes such as accuracy, recall and F1 value, and is able to effectively capture the complex semantic relations in the text and accurately judge the consistency between the script and the star's image, with strong generalization ability. In summary, this study provides a scientific decision support tool for drama selection in the entertainment industry, effectively improves the efficiency and success rate of drama selection, and reduces the risk caused by the inconsistency between the script and the star's image, which is an important application value and theoretical contribution to the management of the star's image and the selection of scripts in the entertainment industry.

Keywords

BERT, NLP, Acting stars, Script selection, Consistency assessment