Objective
Aim of the course is to get an idea and working knowledge of how to deal with spatio-temporal tagged data records within large data sets by discussing and demonstrating approaches that effectively and efficiently support data mining of spatio-temporal patterns.
Description
Data-records almost always contain information about events related to objects/persons that include time and location attributes. These attributes can be used to organize, index and visualize data in search of patterns. Each week possible approaches will be discussed leading to proposed methods that eventually are to be tested on one of two large datasets available for this course. Both datasets count about 10 million records, both are about 19th-Century historic data (civil records: marriage-, birth- and death-certificates 1811-1922, of the Netherlands and library descriptions of all the books in the world published between 1800-1914).
Method
Active participants only: after the first introductory session, each weekly session must be prepared by either reading a paper supplied, or by working out an assignment given out or chosen during the previous session.
Examination
no written exam. Paper presentations and programming assignments
Prerequisites
bachelor CS
Literature
will be provided by the lecturer
Material
will be provided by the lecturer
Remarks
Before students can use subsets of these datasets they have to sign a contract agreeing that all of the supplied data will be kept confidential and used solely for the purpose of trying out and demonstrating methods developed during this seminar.