Mining Weighted Frequent Patterns from Uncertain Data Streams

Abstract

In recent years, data mining has become one of the most demanding areas of computer science. For instance, it helps to discover frequent patterns by applying intelligence tools, techniques and methodologies from various kinds of databases. However, with the rapid growth of modern technology, high volumes of data with different characteristics are generated by many applications. Situation has become much more challenging and sophisticated when datasets are uncertain in nature and flowing at high velocity. Many applications also demand real-time analysis of data depending on current characteristics. Several researches have been made to mitigate the challenges regarding uncertain data streams. However, as the datasets are streaming in nature, frequent patterns of those datasets may be huge in size, and thus may require further mining to find the interesting patterns. Interestingness of patterns can be measured by associating weight with each item. In this paper, we propose a novel tree-based approach called WFPMUDS (Weighted Frequent Patterns mining from Uncertain Data Streams)-growth, which is capable of capturing recent behavior of uncertain data streams and only produces significant (weighted) patterns.