Computational Advertising Research from the Perspective of Psychological Synergy: Review and Outlook

Liu Tong; Huang Minxue; Yu Zhengdong

doi:doi:10.16538/j.cnki.fem.20220429.303

Summary

Computational advertising is defined as a broad, data-driven advertising approach relying on or facilitated by enhanced computing capabilities, mathematical models / algorithms, and the technology infrastructure to create and deliver messages and monitor / surveil individual behavior. Computational advertising has become the mainstream form of Internet advertising due to its characteristics of automation and accuracy. However, the current computational advertising system is facing the challenges of data collection and algorithm selection, which highly impacts the advertising effect. For example, the data paradox (Due to the mobility of consumer behavior, the more the consumer data collected, the more accurate the consumer portrait, but the more complex the calculation, the more serious the privacy problem) and short-sighted decision-making (The advertising content is too changeable to adapt to the scene of consumer behavior, but ignores the long-term construction of brand image) of computational advertising hinder the further improvement of advertising effectiveness. In order to answer this question, this study explores the possible direction of solving the problem based on analyzing the beginning of the problem.

Through combing the relevant literature and industry experience, this study analyzes the emergence and development of computational advertising, obtains the causes of the problems, and verifies them through the method of literature measurement. The results show that the computational advertising model has changed from the previous passive “stimulation – organism – response” model to the active “prediction – matching – iteration” model. The computational advertising system relies on the correlation between data to make synergy decisions, and the data with algorithms become the focal point of advertising studies. The measurement of literature supports this conclusion. Therefore, this study believes that the key to the above problem lies in the conflict between advertising objectives and advertising modes. The historical behavior data of consumers based on the data synergy decision-making model is fragmented and contextual, so the algorithm relies too much on the correlation of data and lacks the ability to explain behavior, and cannot predict and explain consumer behavior. Then, this study explores the possible direction and methods of the future development of computational advertising from the psychological level, constructs the knowledge graph of “User – Product– Scene” with ternary interaction, builds a psychological synergy framework based on the current computational advertising system, introduces the causal inference ability into the decision-making process, and guides the data synergy mode to the psychological synergy mode, so as to predict consumer demand more effectively and stably.

This study makes the following contributions: First, it verifies the data synergy decision-making model of computational advertising. Second, it points out that the current development dilemma faced by computational advertising comes from the conflict between the goal of advertising accuracy and the means of relying on algorithms, and supplementing the insight into consumers’ psychology can effectively fill this loophole. Third, based on the current data synergy decision-making model of computational advertising, it attempts to build a psychological synergy framework, and uses the combination of knowledge graph and algorithms to create the operation basis of psychological synergy model. Fourth, it puts forward the possible direction of the development of computational advertising, that is, introducing new knowledge to have a deeper understanding of consumer behavior, combining it with automatic decision-making, changing from data synergy mode to psychological synergy mode, and changing from computational intelligence mode relying on data and algorithms to cognitive intelligence mode with the ability of understanding and interpretation.