Top-K Miner: top-K identical frequent itemsets discovery without user support threshold

Simple item page
dc.contributor.authorAshraf, Jawad
dc.contributor.authorRehman, Saif ur
dc.contributor.authorHabib, Asad
dc.contributor.authorSalam, Abdus
dc.date.acceptance2015-11-24
dc.date.accessioned2024-10-29T13:32:00Z
dc.date.available2024-10-29T13:32:00Z
dc.date.issued2015-12-31
dc.description.abstractFrequent itemsets (FIs) mining is a prime research area in association rule mining. The customary techniques find FIs or its variants on the basis of either support threshold value or by setting two generic parameters, i.e., N (topmost itemsets) and Kmax (size of the itemsets). However, users are unable to mine the absolute desired number of patterns because they tune these approaches with their approximate parameters settings. We proposed a novel technique, top-K Miner that does not require setting of support threshold, N and Kmax values. Top-K Miner requires the user to specify only a single parameter, i.e., K to find the desired number of frequent patterns called identical frequent itemsets (IFIs). Top-K Miner uses a novel candidate production algorithm called join-FI algorithm. This algorithm uses frequent 2-itemsets to yield one or more candidate itemsets of arbitrary size. The join-FI algorithm follows bottom-up recursive technique to construct candidate-itemsets-search tree. Finally, the generated candidate itemsets are manipulated by the Maintain-Top-K_List algorithm to produce Top-K_List of the IFIs. The proposed top-K Miner algorithm significantly outperforms the generic benchmark techniques even when they are running with the ideal parameters settings.
dc.funderNo external funder
dc.identifier10.1007/s10115-015-0907-7
dc.identifier.citationSaif-Ur-Rehman, Ashraf, J., Habib, A. et al. (2016) Top-K Miner: top-K identical frequent itemsets discovery without user support threshold. Knowledge and Information Systems, 48, pp. 741–762
dc.identifier.doihttps://doi.org/10.1007/s10115-015-0907-7
dc.identifier.issn02191377
dc.identifier.urihttps://hdl.handle.net/2086/24410
dc.peerreviewedYes
dc.publisherSpringer
dc.relation.ispartofKnowledge and Information Systems
dc.relation.ispartofseriesKnowledge and Information Systems
dc.subjectData Mining
dc.titleTop-K Miner: top-K identical frequent itemsets discovery without user support threshold
dc.typeArticle
oaire.citation.issue3
oaire.citation.volume48

Files

License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
4.2 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

School of Computer Science and Informatics