Implementation Projects
Each student should attempt to implement one of the following
projects, or other projects, related to problems in knowledge
discovery discussed in the class.
Every student needs to do a project associated with the class
material. The project can be of the following nature:
- Improvements on existing algorithms
- Combination of existing algorithms
- New scalable algorithms for mining, data cube implementations,
indexing, dimension deduction, etc.
- Integration of knowledge discovery and statistic techniques
Suggested Projects
- Writing educational Java applets to illustrate data mining and data
warehousing concepts (classification trees, clustering, etc.).
- Implementing a visualization module to visualize datacube data (or data
mining results) on the projection screens of the VizRoom (in 3D)
- Implementing an incremental, a distributed or parallel mining algorithm (many possibilities)
- Implementing a similarity search in time series data
- Implementing OLAM algorithms (using data cubes as input) (many possibilities)
- Implementing multi-level association rule mining
- Implementing classification algorithms (many possibilities)
- Implementing clustering algorithms (many possibilities)
- Implementing Discretization of numerical attributes and concept
hierarchy building.
- Extracting interesting metadata about web documents.
- Finding authoritative web documents (such as CLEVER)
- Implementing clustering of Web documents according to document attributes,
usage and content.
- Implementing transaction and sequence identification in web access
logs.
- Implementing association rule extraction from web access logs.
- Implementing web document restructuring.
- Implementing Web document browsing a la OLAP using existing ontologies.
- Designing a query language for querying interesting rules.
- Implementing clustering of access patterns in web access logs.
- Text mining from text documents (e-mails, memos, on-line news, etc.)
- Implementing association rule extraction from images
- Implementing visualization techniques for displaying data mining
results or for helping interactive data mining.
|
