Implementation of 3 similar papers targeting hawkes processes in information diffusion in social networks and comparison of their performance of absolute percentage accuracy (APE) for predicting the final number of retweets with different tracking times of the retweet sequence (10, 20, 30minutes).
This has to be done with the python "tick"([login to view URL]) library which was designed for hawkes processes.
It is necessary to adapt the library to the specific hawkes formulations used in SEISMIC ([login to view URL]), TIDEH ([login to view URL]) and MASEPTIDE ([login to view URL]). The baseline mu_t function of the hawkes process and the self exciting kernel phi should be possible to can be costumized.
The estimation procedure of the parameters should be maximum likelihood estimation (is available in tick but I think it has to be adapted).