DEBI PhD Candidacy Presentation by Mr Bui Dang
DEBI Institute is pleased to invite you to a PhD Candidacy presentation by Mr Bui Dang.
Mining Complex-Structured Process Logs: Enhanced Methods And Applications
Mr Bui Dang.
Date & Venue
Date: 28 November 2011,
Time: 2.00pm - 3.00pm
Venue: DEBII Board Room, Enterprise Unit 4, Technology Park, De Laeter Way, Bentley
Supervisor: Dr Vidy Potdar
Co-supervisors: Dr Fedja Hadzic
Chairperson: Dr Michael Hecker
Business process management is a driving force in improving efficiency and reducing cost in many large organizations. Today, business processes and activities are mostly controlled by computing systems. Process-aware information systems provide detailed information about the activities that have been executed in an event log. This log can be analyzed in different ways to get insights about the process models, the performance bottleneck or other interesting characteristics of the organization. However, the information contained in each log is quite large (thousands to millions of records) and complex, thus requiring sophisticated and efficient data analysis methods.
Various data mining and simulation based algorithms have been tried out in this field in the last decade, nevertheless they mainly focus on the process discovery and conformance checking tasks with limited work on outlier detection and analysis. Furthermore, the event logs are increasingly being represented in semi-structured format using XML based templates to enrich the information content and represent it in a domain oriented way. However, the commonly used XML mining techniques such as frequent subtree mining, and closed/maximal subtree mining have not been explored.
In this research, we investigate the application of frequent subtree mining techniques to discover associations among semi-structured data objects, as well as methods for XML document clustering, outlier detection/analysis/prediction and classification methods that take the structural information into account. The proposed framework will enable mining of semi-structured or tree-structured event logs to discover knowledge patterns capturing interesting information about a broad range of organizational aspects thereby satisfying a wide variety of application tasks. The science and engineering research approach is utilized in this research and evaluation will be performed using real event logs from the industry partner, as well as publicly available real-world and synthetic event log data represented using semi-structured (XML) format.