Optimizing time slots in scientific meetings: a Latent Dirichlet allocation approach

Luca Frigau

2022-01-01

---

Abstract

When participating in scientific conferences, often it happens that sessions with similar topics are scheduled at the same time, thus leading to having to choose which one to follow and giving up the others. Recently, to overcome this problem, an algorithm has been proposed that uses Latent Dirichlet allocation to optimize the allocation of sessions in slots. The results obtained on the Joint Statistical Meetings 2020 program, which concerned more than 40 parallel sessions, have been very interesting. In this paper, we investigate the actual adaptability and effectiveness of this algorithm also for medium-sized conference programs such as that of the SIS.