170 likes | 350 Views
Teaching and Assessing a Data Warehouse Design Course. Karen C. Davis Sandipto Banerjee University of Cincinnati Cincinnati, OH USA TLAD Workshop June 2, 2007. Overview. course structure assessment techniques scope of the course software tool for schema modeling and evolution
E N D
Teaching and Assessing aData Warehouse Design Course Karen C. Davis Sandipto Banerjee University of Cincinnati Cincinnati, OH USA TLAD Workshop June 2, 2007
Overview • course structure • assessment techniques • scope of the course • software tool for schema modeling and evolution • conclusions and future work
Course Mechanics • review of data warehouse design research issues • independent student research • course elements: • class participation/homework: 50% • presentation: 20% • paper: 30% • homework: primarily reading assignments and 3 written discussion points • citations • thoughts (in context) • 1-5 sentences per point • outcome: develop scholarly thinking and writing skills
Expectations for Student Learning • content area knowledge • data warehouse design research • research skills development • oral and written communication • literature survey with domain-specific critical thinking skills • citation formatting
Research Skills Development Activities * draft is marked up but not graded (estimated grades added in 2007)
Peer Evaluation • anonymous feedback given to student • peer scores contribute to homework grade • evaluator receives participation grade • professor score is presentation grade • effective for engaging listeners
Sample Feedback Given to a Student overall score (peer rating): 9.2/10 summary of comments on how to improve the presentation: Nice loud speaking voice, good animation about the topic. The example was confusing, though. You should not try to equate web search engines with multidimensional data (not really the same model or the same kinds of search challenges/goals.) A Google search is over semistructured data and a multidimensional query is over highly structured data. Pay attention to the slide limit! The slides are very cluttered. With practice, the presenter will be a very good speaker. The basic definitions of some terminology are needed. Speak slower and give more comparison between papers. Sentences on slides were too long; sometimes when explaining one idea was not completed before quickly switching to the next sentence. Limit the text you put on a slide; pick the most important information. Presentation was well-prepared and well-presented but there was too much detail on the slides. Don’t talk so fast, concentrate on speaking clearly. Good job on not hiding behind the computer. Good presentation but 24 slides in 15 minutes is 1.6 slides per minute. Talked a bit fast.
Textbooks vs. Papers Indexing in Data Warehouses: Bitmaps and Beyond Karen C. Davis Ashima Gupta
Data Warehousing at the CrossroadsPerspectives WorkshopSchloss Dagstuhl International Conference and Research Center for Computer Science • emerging applications • novel environments • modeling and design • architectures • query processing • ETL • 45 page report • 18 co-authors • 187 references
Student Background Knowledge multiple project experience moderate project experience some practice reading knowledge no experience
select 3 papers on a related topic should be by different research groups at least one should be from the last two years sample topics: data warehousing and XML multidimensional normal forms schema evolution ETL modeling and optimization spatial indexing Starting Independent Research • format: • overview of area with motivating example • summary of each paper’s focus and contribution • compare/contrast approaches • conclusions and future work
DW Conceptual Design and Evolution Tool • creation and population of conceptual data warehouse schema • exploration of the impact of schema evolution • supports core DW features • uses ULD for formal definition of constructs and correctness constraints • uses MDD approach as basis for implementation • SQL Server 2000 • correctness constraints implemented as triggers • evolution operators implemented as stored procedures that utilise triggers
Core Features • fact • measure • dimension • cube • m:1 fact-dimension • 1:m hierarchy • single path
Constraints and Schema Evolution Operators:Triggers and Stored Procedures add evol operator screen shot
Advanced Hierarchy Semantics non-strict non-onto multiple path non-covering
Contributions • structure, organization, and assessment of an introductory data warehouse design course • extensive bibliography (course and student research) • tool for future use in course projects