Association Rules Mining IN High Speed Data Stream Using Sliding Window With the idea of supplementing the input data

Nowadays high volumes of data are stored in database. To identify many of hidden patterns in these data, we need new instrument and techniques. In this regard data mining is introduced as a novel science for searching beneficial patterns from high volume of data. Discovery of association rules is one of most useful patterns that extracted by data mining and it means finding all presented rules in observing the items of data collections so that these rules cover considerable volume of data. The most important phase for finding association rules is finding most repeated patterns. Growing trend of data flows and its appearance in wide range of advanced functions such as communication networks, telecommunication and money laundering resulted in studying of most repeated items in data flows. Whereas despite static database data in the shape of fast and continuous flows. Data mining encounters too many problems. One scan feature, need to unlimited memory and high rate of data entrance are such these problems. Because of these limitations, we follow a sub-collection instead of finding all most repeated patterns in New Year’s which will be most attractive. In this paper we introduce special type of most repeated patterns by highest length in data flow with the idea of complementing entrance data. So far this topic is not investigated in data flow. New algorithm based on sliding window for its searching introduced and performed evaluations are in comparison with other present algorithms in this field.

کلیدواژه ها:

Data flow ، association rule discovery ، sliding window ، most repeated patterns by highest length ، complementing entrance data