Re: sequential pattern mining
Date: June 24, 2020 09:27AM
Hi,
I see.
Then, I think it can be appropriate for sequential pattern mining. In sequential pattern mining the input is a sequence database (a set of sequences) and the goal is to find subsequences that appear frequently in these sequences.
In your case, the sequence database could contain a set of sequences of clicks performed by different experts. Then by applying a sequential pattern mining algorithm, you would find some sequences of clicks that are common to several experts. For this you could use a sequential pattern mining algorithm like CM-SPAM for example. It will also let you specify some constraints such as the max gap constraints (the gap that will be allowed between clicks - do you want for example to find only consecutive clicks or to also skip some clicks?).
If you have a single expert rather than many experts, then you could check episode mining algorithm. Here rather than finding patterns that are common to multiple sequences, the input is a single very long sequence of clicks and you would try to find some subsequence of clicks that appear many times in that long sequences. There are a few algorithms for this in SPMF like TKE, EMMA, etc.
Hope this helps.
Philippe