Speaker
Description
I have been working on open source software (Hoiho) that automatically learns regular expressions that extract features from router hostnames. The general idea is to use a training set with labels inferred using heuristic algorithms, and then learn regular expressions that extract information congruent with those labels. Currently, the software extracts "router names" (portions of a hostname that are in common across all interfaces on a router and unique to that router) and "autonomous system numbers" (portions of a hostname that identify the ASN that operates the router). I have work ongoing to extract geolocation and AS name annotations, and beyond this there might be ideas in the DNS-OARC community on how to apply the idea to other types of hostnames.
