Syllabus for CSci 8715, Spatial Databases, Fall 2007

Monday and Wednesday 04:00 P.M. - 05:15 P.M., EE/Csci 3-111
Instructor and TAs
Role: Name Office & Hours Phone Email
Instructor: Prof. Shashi Shekhar EE/CS 5-203,
Mon: 12:30P.M-1:30P.M, Wed: 3:00P.M.-4:00P.M.
624-8307 shekhar@cs.umn.edu
TA: Mete Celik EE/CS 5-202, Mon,Wed: 3:00P.M.-4:00P.M. 626-7703 mcelik@cs.umn.edu
Class Web Site: http://www.spatial.cs.umn.edu/Courses/Fall07/8715/
Schedule: lecture, homework and examination schedule
Web Pages:
NON-LOCAL
  • Search: Books (amazon.com), Papers (DBLP), CiteSeer, Google Scholar
  • Browse or download papers from Journals: IEEE TKDE, GeoInformatica Journal, IJGIS, ACM TODS, IEEE TGRS
  • Conference Proceedings: CIKM/ACMGIS, SIGKDD, SSD, GIScience, W2GIS, VLDB, SIGMOD
  • Bulletins: IEEE DE Bulletin, ACM SIGMOD Bulletin, ACM SIGKDD Explorations
    LOCAL
  • Class notes, TA Announcements, HW Feedbacks, Project-list-1, Project-list-2 (Other resources for project) , Group info.
  • Midterm Exam Sample ('04), Midterm Exam Sample ('01), Sample Question
  • Sample Proposal 1, Sample Proposal 2
  • Sample Project Report 1 , Sample Project Report 2 , Sample Project Report 3
    Pre-requisite: Familiarity with Relational Databases or Geographic Information Systems.
    Text Book: Spatial Databases: A Tour, S. Shekhar and S. Chawla, Prentice Hall, 2003, ISBN 013-017480-7 .
    Supplementary Material: A collection of papers.
    Topics:
    1. Application Domains of Geographical Information Systems (GIS), Common GIS data types and analysis.
    2. Conceptual Data Models for spatial databases (e.g. pictogram enhanced ERDs).
    3. Logical data models for spatial databases: rastor model (map algebra), vector model (OGIS/ SQL1999).
    4. Physical data models for spatial databases: Clustering methods (space filling curves), Storage methods (R-tree, Grid files), Concurrency control (R-link trees), Compression methods for rastor and vector data.
    5. Query Optimization: strategies for range query, nearest neighbor query, spatial joins (e.g. tree matching), cost models for new strategies, impact on rule based optimization.
    6. Spatial networks: Conceptual, logical, and physical data models, query languates, graph algorithms, access methods.
    7. Mining spatial database: spatial auto-correlation, co-location patterns, spatial outliers, classification methods (SAR, MRF).
    8. Rastor databases: Raster image operations, content-based retrieval, spatial data warehouses.
    9. Trends: mobile wireless applications, spatio-temporal data, security, geo-sensor network.

    Examinations and Assignments: The main objective of this class is to study research methods and literature in spatial database systems. Core research skills of literature analysis, innovation, evaluation of new ideas, and communication are emphasized via homeworks and projects.

    Various acivities in a research seminar courses are linked to the goals of the audience. Many students may like to get a broad overview of the research topics, methodologies, major results, open problems and potential future directions. In-class written examinations on survey papers from the reading list will be useful towards this purpose.

    Ph.D. students in this course may benefit from an examination similar to the take-home examination in the Written Preliminary Examination. A take-home examination analyzing a research paper (potentialy from outside the reading list) will help here. Potential sources for the paper would be conference proceedings or journals such as those listed above.

    Honors undergradaute students as well as M.S. students in the course may benefit from projects and term papers similar to those for their thesis requirements. A project broken down in several steps will be relevant here.

    Cheating/ Collaboration: Getting help from services like general debugging service (GDS), buying term papers from web-sites (e.g. cheaters.com), copying someone else's assignment, or the common solution of written or programming assignments will be considered cheating. The purpose of assignments is to provide individual feedback as well to get you thinking. Interaction for the purpose of understanding a problem is not considered cheating and will be encouraged. However, the actual solution to problems must be one's own.

    Helpful Comments: This class is Very Interesting and Useful for audience interested in database systems research as well as in honors/Master/Doctoral projects. We will explore a number of current research areas which are very important yet fairly open for research. Spatial databases continue to be the heart of information management in areas ranging from business (e.g. google earth, navigation device) to scientific domains (e.g. earth observation systems, epidemiology).

    To get full benefit out of the class you have to work independently and regularly. Read the papers before the meeting and bring comments for discussion. Plan to spend at least 6 hrs a week (a little more during first few weeks till you feel comfortable with geographic information and queries) on this class doing projects or reading.