Computational Analysis of Pluralistic Ignorance in Shaping Corporate Employment Image Using the EPO-Based SINN Model on Social Media

With the rise of social media, corporate employment image on social media platforms critically impacts the employment situation. Recently, there have been evident differences in the corporate employment image of different industries on Chinese social media platforms. Pluralistic ignorance widely exists in online discussions of different themes and may also influence people’s opinion of corporate employment image. However, little is known about the effect manner and degree of pluralistic ignorance in discussing different corporate employment images. This paper integrates the latest opinion dynamics model, expressed-private-opinions (EPO), and deep learning through an EPO-based SINN and applies the proposed model to a real-world dataset from Chinese social media. We focus on five important industries in China as research objects and constructed datasets containing 110,763 pieces of text data and then compare the performance of the proposed method with baseline methods on the datasets. EPO-based SINN gets the F1-scores of 0.801, 0.781, 0.744, 0.751 and 0.795 on five industries’ datasets and outperforms baseline methods. The prediction results confirm the existence of pluralistic ignorance, and the optimized resilience parameter in EPO-based SINN reveals the different scales of pluralistic ignorance in the five themes. Furthermore, our quantitative analysis shows that the proportion and polar distribution of zealots in social networks positively impact the scale of pluralistic ignorance. Our work reveals the mechanisms between pluralistic ignorance and corporate employment image and infers the main driver making a difference in the scale of pluralistic ignorance across discussions of different industries.