Case-Based Reasoning and Attribute Features Mining for Posting-Popularity Prediction: A Case Study in the Online Automobile Community

Abstract

Social media is in a dynamic environment of real-time interaction, and users generate overwhelming and high-dimensional information at all times. A new case-based reasoning (CBR) method combined with attribute features mining for posting-popularity prediction in online communities is explored from the perspective of imitating human knowledge reasoning in artificial intelligence. To improve the quality of algorithms for CBR approach retrieval and extraction and describe high-dimensional network information in the form of the CBR case, the idea of intrinsically interpretable attribute features is proposed. Based on the theory and research of the social network combined with computer technology of data analysis and text mining, useful information could be successfully collected from massive network information, from which the simple information features and covered information features are summarized and extracted to explain the popularity of the online automobile community. We convert complex network information into a set of interpretable attribute features of different data types and construct the CBR approach presentation system of network postings. Moreover, this paper constructs the network posting cases database suitable for the social media network environment. To deal with extreme situations caused by network application scenarios, trimming suggestions and methods for similar posting cases of the network community have been provided. The case study shows that the developed posting popularity prediction method is suitable for the complex social network environment and can effectively support decision makers to fully use the experience and knowledge of historical cases and find an excellent solution to forecasting popularity in the network community.