Key Research Highlights > Databases & Big Data Analytics > Integrated Mining
and Visualisation of Complex Data
Wynne HSU Mong Li LEE |
Integrated Mining and Visualisation of Complex Data
Objective Most discovery algorithms used to identify data gold nuggets focus on specific data types. Modern applications, however, produce copious amounts of diverse data types. We want to be able to handle datasets with a variety of data types in an efficient manner. Results We have met the objectives of this project through the development of a framework to perform integrated data mining of big data with diverse data types. The framework consists of algorithms for mining patterns from interval-based events, lag patterns involving motifs in time series data, spatial interaction patterns. Additionally, we have designed algorithms to discover duration-aware region rules and path rules from trajectories, which can able to capture associations among different complex data types and demonstrate how these patterns can be used to improve the classification accuracy in various real world datasets. Furthermore, we developed a a tool to generate and highlight interesting patterns discovered from different data types. The tool allows the visualisation of event incidences, clusters and heat maps. Ongoing research aims to develop an interactive system for the visualisation and analysis of trajectories |