SWIM (Sliding Window Incremental Miner).
Swim is a verification-based approach to the popular frequent pattern mining
problem, where by verification we mean conditional counting. You can find the
main paper introducing SWIM and its fast verifiers
here. In addition
to the stream mining scenario, as shown in the aforementioned paper, we have
found other potential applications for our verifiers, including stream
monitoring, privacy preserving association rule mining, enhancing traditional
static frequent pattern mining. You may find other benefits of having a fast
conditional counter (=verifier) in your own application. Therefore, we have made
our implementations available online. You can obtain the
latest implementation of SWIM
(and its underlying verifiers), as long as it is solely used for academic
purposes (e.g. comparing against your own algorithm, benchmark, boosting your
own approach, etc), and you mention
the original source.
