CSE 291, Topics in Database Systems: Information Integration, OLAP and Data Warehousing, Multimedia Information Systems
| Course Topics | Prerequisites | Grading |
Schedule Monday, Wednesday 11:30am.
Instructor Yannis Papakonstantinou
During the last five years, changes in the information landscape have posed new challenges and requirements on database systems. CSE 291, "Topics in Database Systems", will describe the commercial practice, recent research advances, and the challenges and opportunities in the areas of information integration, OLAP / data warehousing, and support for multimedia objects.
The following is a tentative syllabus of the topics that will be covered by the course. The exact syllabus will also depend on students interests.
| OLAP and Data Warehousing | Information Integration | Multimedia |
Information analysts (typically managers and executives) require on-line responses to operations such as summarizing data, rolling up and down in the summary hierarchy, analyzing what-if scenarios, and so on. In a nutshell, imagine Excel running on workbooks larger than 10 Gbytes. Brute force data warehousing and On Line Analytical Processing (OLAP) tools meet the on-line requirement by collecting off line data from the (potentially multiple and heterogeneous) sources and precomputing a large number of views and indexes. At the same time database research promises a new generation of more efficient and flexible warehouses.
The course first covers the essentials of data warehousing and OLAP technology:
The rapid growth (largely due to the Internet) of data available by distributed autonomous sources overwhelms information consumers. Mediators assist the users by providing them a single integrated view of the distributed and heterogeneous data, and by allowing them to query this information using high level query languages - as in database systems. During run time the mediator decomposes the user queries into queries that are sent to the sources and collect the necessary information.
However, query processing in mediator systems is significantly different than query processing in a distributed database. The sources typically provide limited interfaces, they export semistructured data, redundant and conflicting data are commonplace, and query optimization should focus on response time (for browsing applications.) We’ll describe a series of techniques and algorithms suggested for handling the problems above and we’ll investigate open problems.
Including multimedia in existing databases requires more than enough disk space. Multimedia queries ask for ranked results, e.g., rank the x-rays images by order of similarity with a given image. Furthermore, primary key based indexing, or even spatial indexing methods (such as R-trees), will not work for nearest neighbor searches in spaces comprised of one hundred (or even more) dimensions. We will investigate issues both on the semantics as well as the implementation of querying multimedia systems.
Essential knowledge about data structures, searching, and database systems is required. Whenever the knowledge of more advanced issues is required it will be covered by lectures and handouts. In particular, the students should be acquainted with:
The course will be given with letter grade and pass/fail options. The evaluation will be based on class participation and one of the following:
