Re: motivation for top-k closed high utility itemsets mining
Date: November 06, 2019 03:29PM
Yes, I think there is no paper on that topic by doing a very quick search on Google Scholar.
Yes, I think this would be a useful topic. The motivation of top-k high utility itemset mining is that we dont want to set the min utility threshold and instead we want to find the top k itemsets that yield the highest profit. But these itemsets may contain redundancy... So if you mine the top-k closed high utility itemset, you would find the k most profitable itemsets that are non redundant. I think this is definitely useful.
Now, in terms of challenges, I think that doing top-k closed high utility itemset mining is not so straightforward. So I think it would also be a challenging topic, perhaps, which should thus be interesting.
To do this project, I would suggest to look at:
- the papers on high utility itemset mining, closed high utility itemset mining and top-k high utility itemset mining, and you can find some code in SPMF and datasets.
- the papers bout top-k closed frequent itemset mining... i think there should be some.. so this would give you some ideas about how to combine top-k with itemset mining.
Then you could see how to put all this together.