Brief Announcement: Weighted Partial Message Matching for Implicit Multicast Systems

In implicit multicast, receiving processes delineate the messages they wish to receive by specifying predicates, also called filters, on the message's content. In weighted partial message matching, distributed applications using implicit multicast protocols do not require that a message match all the elementary constraints constituting the filter. Typically, receiving processes in such applications associate a non-negative weight to each constraint and require that the match score, i.e., sum of the weights of matching constraints, exceeds a threshold value. In this paper, we consider top-k weighted partial matching, where a process is interested in multicasting a message only to k<n other processes corresponding to the top-k match scores. This is a fundamental problem underlying online advertising platforms, mobile social networks, online dating, etc.