A Study on the Impact of Algorithmic Bias on the Fairness of Film and Television Communication and Regulation

Abstract

Algorithmic bias is the bias of the algorithm in the operation process, resulting in unfair and unreasonable repeatable results, this paper starts from the perspective of algorithmic recommendation to explore the impact of algorithmic bias on the fairness of film and television dissemination. Using video feature data and user behavior data of short video film and television area as research objects, identifying the fairness of film and television dissemination, selecting feature variables of algorithmic bias, constructing a multiple linear regression model of the influence of algorithmic bias on the fairness of film and television dissemination, and introducing SHAP interpretation framework to quantify and attribute the importance of each feature variable. The results found that recommendation diversity, equalization of opportunities, and population parity have significant positive impacts on the fairness of film and television communication, and the exposure inequality index and discovery path bias both have significant negative impacts on the fairness of film and television communication, which are all significant at the 5% level, and their impacts on the fairness of film and television communication are different in terms of the way, direction, and strength of their impacts. Combined with the advanced regulation experience of today's society and various social media platforms, suggestions and countermeasures are proposed for the regulation direction of algorithmic recommendation.