Hello Prof. Philippe,

I am interested in sequential pattern mining with multiple supports because I need to obtain infrequent patterns that may contain important information and that is lost in traditional sequential pattern mining algorithms.

I would like to ask you if there exist any algorithm in SPMF that mines sequential data with multiple supports. I would also like to know your opinion regarding this topic please: does it enormously increase the complexity of the algorithm, what is the best algorithm performing this type of mining, and do you think that Hirate and Yamana algorithm could be feasibly updated by having multiple supports?

Thank you for your help,
jbd

Hi,

Currently, there is no algorithm in SPMF for sequential pattern mining with multiple thresholds but there are some for itemset mining.

Transforming a pattern mining algorithm to handle multiple threshold is usually not so difficult. It would require to read how people have done it in frequent sequential pattern mining and then copy the idea and integrate it to Hirate and Yamana.

I am not familiar with the paper on sequential pattern mining with multiple thresholds but I know that for itemset mining, it is not difficult to do. It just require to use some special tricks such as sorting items according to some order, and then multiple thresholds can be used. I assume that doing it for sequential pattern mining would required to use a similar strategy and thus that it could be adapted for Hirate & Yamana.

But to make sure, it would require to read the papers. Actually, this topic of adding multiple thresholds in Hirate and Yamana could even perhaps be the topic of a short research paper if there is some new challenges, but I am not sure about that (perhaps that it just require a combination of existing ideas rather than some new idea).

Best,

Philippe



Edited 1 time(s). Last edit at 02/03/2019 05:14PM by webmasterphilfv.

Thank you for your detailed response.
jbd