- australiana
About Me > Research > Temporal Data Mining

Temporal Data Mining

Temporal data mining deals with the problem of mining knowledge from temporal data, which can be either symbolic sequences or numerical time series. Both types of temporal data have been intensively studied in data mining and statistics.

Previous Project

Project Title

Development of Methods for Storing, Accessing, and Visualizing Temporal Patterns for Selecting the Best Patterns.


In  this  era  of  information,  one  of  the  major  problems  frequently  encountered  by
companies is how to process their data into information that provides benefits to the  company. In order to face this challenge many techniques  have been developed to find the  information  from  the  data.  This  information  is  usually  presented  in  the  form  of patterns or rules. However, most of these techniques focus on the discovery of non-temporal patterns from non-temporal data. In reality, much of the data owned by the company is in the form of temporal data. Although several methods for discovering patterns from temporal data have been developed these techniques are still weak in terms of post-processing the discoverd patterns. In order to discover the best pattern, post-processing role is very important because from thousands of patterns generated by data mining algorithms only a small part of them is useful, and this can only be identified through post-processing.
Selecting the pattern from thousands of temporal patterns is not easy work. Because of that, in this research we develop a mechanism that can facilitate users in selecting the best temporal patterns, especially sequential patterns. Our research has developed a method for storing and accessing sequential patterns efficiently. It has also developed a visualization method that allows a group of sequential pattern to be displayed in the form of tree that can be easily used by the user for selecting the pattern.

This project is supported by PENELITIAN HIBAH MULTI TAHUN (Penelitian Fundamental 2009 and 2010).