I am researching some concepts of practical 2 party computation and was curious if anyone could point me to a research paper that is focused on private set intersection where one party has a very large set and the other has a comparatively small set. Specifically, a user has a small set and the server has a large set. Since the server will handle thousands or millions of requests, I am more concerned about the performance for the server.
The specific use case that I had in mind is contact discovery. Meaning, a user wants to know who else is using a service but does not want to disclose their contacts to the server in plain text. Ideally the server does not learn the outcome of the intersection. At a minimum the server should only learn the size of the users contact list and the results of the intersection.
I found this paper that appears to be a breakthrough in performance but it's benchmarks are based on two sets of equal size. Is there existing research on a PSI that is optimized for non-equal set size?