Abstract: 【Purpose/significance】Consumer behavior information that hidden in product reviews is significant to enhance the competi?
tiveness of enterprises. Through the mining and reasoning analysis of consumer behavior characteristics, enterprises can not only grasp more specific details about their own products and services, but also promote the implementation of scientific business decisions.【Method/process】This paper proposes a competitive intelligence mining method for product online reviews based on event logic graph(ELG), and conducts an empirical study with Huawei mate20 mobile online reviews as the intelligence source. Analyzes the char? acteristics of consumer behavior by mining the causal variables data from the reviews with LTP, LDA Clustering and Gephi, explores the generation mechanism and development law of behavior, and then applies the results to enterprise marketing;【Result/conclusion】The results show that the formation and development of online consumer behavior is a series of rigorous causal logic evolution process, including three stages: pre-purchase demand, purchase decision, and post-purchase evaluation, among which factors that drive con? sumers' final purchase behavior are multidimensional; the experience and process of service acquisition are important factors that af? fect the generation of satisfaction. This method can effectively mine and reveal the characteristics of online consumption behavior and can provide a new competitive intelligence researching method for enterprises.【Innovation/limitation】This paper use Event Logic Graph to analyze online reviews, which can expand the methods of big data mining. But it is very difficult to extraction of rule events with artificial intelligence, because the corresponding database has not been set up