Using SinaWeibo Microblogs to Identify Complaints of Food Customers
Feng Chen, Weiguang Qin · 2022
Monitoring and responding to complaints about food by consumers are critical for enterprises and governments. Here, we design a method to identify and analyze microblogs posted on SinaWeibo. We proposed a machine learning model to classify the microblogs. Our model used multimodal features and improved a term classification weighting method. We used LDA to identify the most common topics in complaints and summarized bloggers' complaint behavior. Our model obtained an optimal weighted F value of 0.942 using only 4 features. We found that linguistic and multimedia features were more important than social features. When expressing complaints, bloggers were more inclined to mention (@) other bloggers, and most bloggers (80.08%) used images or videos. However, the communication effect of complaint microblogs was even weaker than that of noncomplaint microblogs. Our proposed methods were effective social media monitoring and analysis tools. Data mining of SinaWeibo may reduce information asymmetry and enhance the ability of governments to rapidly respond to community concerns and awareness.