Date and time: May 3 2012 (10:30 – 12:30).
Room: 403 NB
10th Anniversary of the Knowledge Engineering Group
- Vojtěch Svátek, KIZI, VŠE Praha
A very short review of the 10 years of both the seminar series and the informal research group tight to it.
LISp-Miner system: present state and further development
- Milan Šimůnek, LISp, VŠE Praha
The LISp-Miner system has been developed at the Faculty of Informatics and Statistics, UEP, since 1995/96, and follows the long-term development of the GUHA method with roots in the 1960s. The main goals of the system are to support the education of KDD, to serve as a platform for research in the area of KDD, and, last but not least, to provide a background engine for other KDD projects (namely, EverMiner and SEWEBAR).
The most recent achievements will be presented: distributed computation of data mining tasks using a PC grid, the LM TaskPooler module for batch processing of many tasks on a single computer, the ETree-Miner procedure mining for so called exploration trees, and, finally, the LM SwbExporter a LM SwbImporter modules for data exchange with other systems (mainly from the SEWEBAR project).
Further directions of the development of LISp-Miner concern its utilization in the EverMiner project (i.e., automation of the KDD process based on domain knowledge and long-term know-how of how a KDD analysis is done manually). A module, currently in design, is Reverse-Miner, aka “data-mining turned inside-out”. The aim is to generate original data when their properties and patterns are given. This is very suitable for preparing educational data for analysis by students, where each student/team should have its own unique version of data, while the teacher possesses perfect knowledge on what \"nuggets\"
are hidden in this particular dataset.