Predictive Model of Cultural Product Search Interest Based on Collective Memory Hawkes Process

In the online world, social media significantly influences users' search interests. It is a common belief that content that is popular on one social media platform is likely to be equally popular on other platforms. However, there is a relative lack of in-depth research on the interactions between different social media platforms and their underlying mechanisms. To address this issue, this study constructs a unique dataset containing search data of Vevo artist names on Google and corresponding music video playback data on YouTube. Based on these data, this study proposes a Hawkes process model incorporating collective memory triggering theory to explore how social media affects users' search interest and further predicts search interest. One interesting finding during the analysis was that the number of views of a YouTube video can predict to some extent an artist's search interest on Google. Specifically, when YouTube video views peak, it can be expected that the artist's search interest on Google will also peak within the next 2 to 3 days. To evaluate the efficacy of the model, this paper conducts a series of experiments on the constructed dataset and compares the predictive performance of the model with other benchmark models. The experimental results show that the collective memory-based Hawkes process model proposed in this paper has significant advantages in predicting users' interest in searching for cultural products.