Exaggerated Advertisement Inspection System for Judging the Suitability of Advertisements in Social Media Environment

Abstract

Recently, as the social media markets are expanding, the amount of health functional food advertisements posted by individual users such as influencers and social media promoters is increasing. Therefore, users need a system that supports them to post false advertisements after inspecting them. In this paper, we propose an exaggerated advertisement inspection system that judges the suitable of advertisements and presents the grounds for disqualification. The proposed system consists of a module that classifies advertisements and explainable artificial intelligence(XAI). The system provides a rationale for judging the results of advertising classification and exaggerated advertisements. Therefore, the user may know why his or her writing is classified as exaggerated advertisement. The language model and embedding model, used in the exaggerated advertisement classification step, check the accuracy of the confusion matrix through the evaluation data. The XAI model checks performance by inputting data designated as exaggerated advertisement by health functional food-related institutions.