Mining Frequent Itemsets (MFI) over Data Streams: Variable Window Size (VWS) by Context Variation Analysis (CVA) of the Streaming Transactions

The challenges with respect to mining frequent items over data streaming engaging variable window size and low memory space are addressed in this research paper. To check the varying point of context change in streaming transaction we have developed a window structure which will be in two levels and...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Reddy, V. Sidda, Rao, Dr. T. V, Govardhan, Dr. A
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The challenges with respect to mining frequent items over data streaming engaging variable window size and low memory space are addressed in this research paper. To check the varying point of context change in streaming transaction we have developed a window structure which will be in two levels and supports in fixing the window size instantly and controls the heterogeneities and assures homogeneities among transactions added to the window. To minimize the memory utilization, computational cost and improve the process scalability, this design will allow fixing the coverage or support at window level. Here in this document, an incremental mining of frequent item-sets from the window and a context variation analysis approach are being introduced. The complete technology that we are presenting in this document is named as Mining Frequent Item-sets using Variable Window Size fixed by Context Variation Analysis (MFI-VWS-CVA). There are clear boundaries among frequent and infrequent item-sets in specific item-sets. In this design we have used window size change to represent the conceptual drift in an information stream. As it were, whenever there is a problem in setting window size effectively the item-set will be infrequent. The experiments that we have executed and documented proved that the algorithm that we have designed is much efficient than that of existing.
DOI:10.48550/arxiv.1408.3175