A Recommendation Algorithm Based on Enthusiasm Function Theory

In most modern e-commerce application scenarios, users' preferences for items change over time. Traditional recommendation algorithms can only construct recommendation models based on users' historical behavior or the current attributes of items, often ignoring the phenomenon that some behaviors of users have delayed correlation, thus affecting the final recommendation effect. This paper investigates the time relationship between the characteristics of consumer goods and consumer behavior by introducing the concept of an enthusiastic degree. Then a recommendation algorithm based on enthusiastic degree function theory is designed, which can predict the time-varying curve of the user's preference for item features and accurately calculate the current time point or even any future time point's enthusiasm for items, thus greatly improving the recommendation quality. Then experiments are carried out in the internal environments of enterprises, and the results show that the proposed recommendation algorithm has better optimization results than the traditional collaborative recommendation algorithm for recommending commodities with time-varying demand. Finally, the future research directions of the project have been outlined.