Projects for Students
The Database stream pursues a project-oriented approach. The essence of the project-oriented approach is that a selected database course of each semester is associated with a mini-project. Students will work in small groups on real world problems (implement an algorithm, test the technique for sythetic and real-world datasets, etc) and write a 10 page report by the end of each semester. The project will be discussed during the oral course exam and constitutes a significant part of the final course exam.
The master degree project consist of 4 mini-projects distributed over the 4 semesters of the M.Sc. program. The DB stream offers master degree projects consisting of coherent sequence of mini-projects. For example, a project can include an implementation of a classical clustering algorithm as the DWDM mini-project, extension of the algorithm for a distributed environment (distributed databases), extension of the clustering algorithm for data with a temporal attribute (temporal and spatial databases), and implementation of the algorithm for streaming data (the 4th semester). The master thesis shall result in a scientific article that is submitted to a conference or journal. The mini-projects can also be completed independently, so that students are not forced to do all 4 miniprojects.
The work load of each mini-project is the following:
- Data Warehouse and Data Mining - 100 hours
- Distributed databases - 50 hours
- Temporal and Spatial Databases - 50 hours (+ 100 hours free choice)
- Final report - 750 hours
The work load for the mini-project of temporal and spatial databases is 50 hours if the student does not choose to do an extended project (free choice), and 150 hours otherwise. In case the student does not choose to do an extended mini-project the additional hours will be substituted with another course.
The DB Stream offers the following projects:
- Sepia String Selectivity
- Temporal database system
- Hierarchical heavy hitters
- Dbscan clustering algorithm
- A multidimensional aggregate-join operator
- Birch clustering algorithm
- Cure clustering algorithm
- Id3 classification algorithm
- Cleaning text databases
- Investigation of Pedigree Data
- Approximate String Mathing in Time, Indepenedent of the Size of the Database
- APDF Method for High Dimensional Data
- CLIQUE Clustering
- Fourier Method for Probability Density Functions
