1. Getting Started Writing a Thesis/Dissertation
Dr. Karen C. Davis
Electrical & Computer Engineering Dept.

2. Graduation

3. ECES 877 Advanced Data Models and Query Optimization
logical
physical
advanced data models
object-relational
data warehouse
XML
Spring 2007
coming to a classroom
near you!

4. Relational Algebra Query Trees
Sujan Turlapaty’s
thesis defense:
Performance Analysis of Self-Maintainable Data Warehousing Algorithms, 11/99

5. Multiple View Processing Plan (MVPP)
⋈orderkey (v7)
Customer (C)
Orders (O)
Lineitem (L)
Nation (N)
Part (P) Q1 Q2 Q3
πO.orderkey, O.shippriority
(v9)
πC.custkey, C.name, C.acctbal, N.name, C.address, C.phone
(v12)
πP.type, L.extendedprice
(v15)
σ C.mktsegment = “building”
and L.shipdate = “ ” (v8)
σ O.orderdate = “ ”
(v11)
σ L.shipdate = “ ”
(v14)
⋈nationkey
(v10)
⋈custkey (v6)
⋈partkey
(v13)
πname, address, phone, acctbal, nationkey, custkey, mktsegment (v1)
πorderkey, orderdate, custkey, shippriority
(v2)
πpartkey, orderkey, shipdate, extendedprice
(v3)
πnationkey, name
(v4)
πpartkey, type(v5)
view chromosome:
index chromosome:
Fitness: sum of query processing costs of individual queries using the views and indexes selected
This is a typical MVPP. We saw this figure previously to understand the huge solution space.This is built by merging the query tress of Q1 , q2 AND q3.A typical view selection solution ,a view chromosome would like this.A 1 indicates that the view is selected and a 0 that it is not.Similarly, in index chromosome, since the indexes can be built on selection and join attrbutes, If the the number of indexes is 7
The index chromosome looks like.
Talk about the correction to the algorithm of moving selections.
thesis defense of Sirisha Machiraju: Space Allocation for Materialized Views and Indexes Using Genetic Algorithms, June 2002

6. BH System Architecture
Michael Brant, Binding Hash Technique for XML Query Optimization, 2006

7. Ph.D. (2)
Satish Venkatesan, 1996, Database Modeling for Electronic Design Automation Environments, awarded ECECS Outstanding Dissertation Award, 1996.
Yunsong Zhan, XML-based Data Integration for Application Interoperability, 2002.
M.S. (24)
Lun Ye, A Compiler Cooperative Dynamic Memory Management System for C++, 1993.
Ron Meade, EasyOpt: A Design Optimization Interface Package, 1994.
Rao Seshagiri Kasinadhuni, Design and Performance Issues of Client-Server DBMS Architectures, 1994.
Samir Nigam, Transformation-based Semantic Query Optimization for Object-Oriented Databases, 1994.
Baskaran Dharmarajan, The Property Map: A Theoretical Foundation and Query Optimization Algorithms, 1997.
Mala Rajamani, Reduction and Maintenance of Self-maintainable Views for Data Warehousing, 1997.
Veena Pandiri, A Global Framework for Distributed Agent-based Systems, 1997.
Radha Ganapathy, Selection of Self-Maintainable Views to Materialize in a Data Warehouse, 1998.
Vishal Sheth, Extended Property Maps: An Efficient Access Mechanism for Retrieval from Large Data Sets, 1998.
Gayathri Krishnan, Physical Schema Design for Object Databases, 1998.
Shobha Ravishankar, Object-Oriented Index Selection and Integration, 1998.
Ji Qin, Access Plan Generation for Property Maps and Multidimensional Indexes, 1999.
Sujan Turlapaty, Performance Analysis of Self-Maintainable Data Warehousing Algorithms, 1999.
Unmi Tina Kang, Path Inherited Dictionary Index (PIDI): An Integrated Object-Oriented Database Index, 2000.
Jennifer Grommon-Litton, Heuristic Design Algorithms and Evaluation Methods for Property Maps, 2000.
Rajeswari Malladi, Applying Multiple Query Optimization in Mobile Databases, 2001.
Xioaming Du, Dynamic Channel and Broadcast Disk Organization in Mobile Databases, 2001.
Krishnamoorthy Janakiraman, Entity Identification Using Data Mining Techniques, 2001.
Casie Phipps, Migrating an Operational Database Schema to Data Warehouse Schemas, 2002.
Ashima Gupta, Performance Comparison of Property Map Indexing and Bitmap Indexing for Data Warehousing, 2002.
Sirisha Machiraju, Space Allocation for Materialized Views and Indexes Using Genetic Algorithms, 2002.
Ravi Darira, A Design Framework for Property Maps, 2006.
Micheale Brant, Binding Hash Technique for XML Query Optimization, 2006.
Janet Rajan, A Framework for Medical Acronym Disambiguation, 2007.
My Students

8. Thesis/Dissertation Organization
title page, abstract, dedication, table of contents, list of figures, list of tables
Introduction
Related Research
Foundations
Results (may be several chapters)
Conclusions and Future Work
Appendices

9. Sample Table of Contents

10. Introduction introduce the general topic area
narrow the focus to specific topic
motivate the research
why is it needed?
who will benefit from the research?
conclude with a clear statement of the problem
give a statement of the work
provide an overview of the thesis (one sentence per chapter)

11. Sample Introduction
conventional database systems are increasingly leveraged for organizational decision-making
analysis systems are different than conventional operational systems because …
… because of these differences, designing
a data warehouse has challenges …
this thesis addresses specific phases of design

12. Research Objectives
general research objective: one sentence describing what you hope to accomplish (not how!)

13. Parallel Sections: Statement of the Work
specific research objectives: partition the general objective into sub-goals
research plan/methodology/tasks/approach: revisit the objectives
your approach to solving the problem
each objective has an associated task or approach to satisfy the objective
expected contributions: revisit the methodology
what will you know or have when you’ve done the task?
potential impact of your work

14. Sample Parallel Sections
specific research objectives accomplish
the general research objective
methodology defines approach to accomplishing
the objectives
expected contributions describe
what will be accomplished by executing
the methodology

15. Related Research focused around your topic; not a tutorial!
compare/contrast to your approach
tables with features/research efforts are concise, readable way to summarize

16. Examples of Summary Tables

17. Foundations work you build on (your own or someone else’s)
definitions, theorems, models, system

18. Research Discuss conventions, setup, hypotheses of experiments, proofs
why did you do it?
what did you learn from it?
Presenting
figures
algorithms
tables
graphs
Sample! Don’t do a dump of everything … put everything in appendices and discuss representative results in the body of the thesis or dissertation

19. Example Experiment Setup
Goals
What are the comparative storage and retrieval cost of REBSI and PMaps in different scenarios?
How is individual and relative performance affected by parameters such as blocksize, database size, selectivity of queries and cardinality of attributes, kind of queries, property ordering?
Can PMaps design and performance be improved using this knowledge?
In what conditions is it better to use either index?
Query Set PMap PMap Performance
[1…6] properties and Storage Cost
Word Size (ws) (pu, pstring)
{16, 32}
Tuple size (t)
{1,000,000, 50,000}
Blocksize (SB)
{2048, 4096, 8192} REBSI Performance
and Storage Cost
Scaling Factor (sf)
{min, …, 10}
PMap Creator
PMap
storage and performance measurement
simulator
REBSI
storage and performance measurement simulator
query set (which fixes the dimensionality of the index, d, and cardinality, c, we have 2 values of ws (word size)
Creation: input to the PMap creation algos determine the PMap properties & pu (pstring utilization)
Query processing: pfilterl, pfilterh, Sq and Sqt for each of the queries

20. Example Presentation of Results
Observations:
REBSI performance improves as the sf becomes larger.
REBSI performance improves as cardinality becomes smaller.
PAvg performance deteriorates as cardinality becomes smaller.
PAvg is better than REBSI min_sf (4) for all queries.
PMin << pages retrieved by any REBSI.
PMap retrieves fewer pages for multi-attribute queries than single attribute queries.
Queries are ordered by the difference of REBSI with min_sf and PAvg.
number figures (e.g., Figure 3.2)
refer to the figures in text
“In Figure 3.2, results for the HCAQS are shown.”
“Figure 3.2 shows HCAQS results.”
explain the conventions
“The x-axis shows individual queries and the y-axis shows index pages retrieved. The queries are ordered by decreasing cardinality.”
offer observations to help the reader see what is important or interesting
“REBSI performance improves as the cardinality decreases.”
discuss possible reasons for the observed results
give general conclusions
x-axis – queries; y-axis - index pages. Low number of pages retrieved for a query indicates good performance
(PMin), maximum (PMax) and average (PAvg) number of index pages shown.
Queries ordered in decreasing order of difference of PAvg and pages retrieved by REBSI with the min_sf.

21. Conclusions and Future Work
revisit objectives
what was accomplished?
what was learned?
topics for future work
extensions
open questions

22. Conclusions: Future Work:
BH method work well for deeply nested queries with few branches (non-bushy)
BH Indexing technique requires further optimization
BindingCollection is a flexible data structure
Can be used in to generate witness trees for processing embedded Xpath expressions
Used to process Xpath expression directly
Can use a different indexing schemes
Future Work:
Modify indexing technique to increase performance and perform inequality matching
Expand Post-order Traversal to support more TAX pattern tree features: e.g., value-based joins
Expand more extensive performance study

23. Citations allow the reader to follow up on the topic
fill in background information
judge what you’ve said by reading original sources
relieve you of the burden of going over all territory on a subject
strengthen/justify your point
respect your peers by acknowledging their contributions
[vL78]

24. Citations not a part of speech!
#11: never, ever, use a bracketed number as if it were the name of an author or a work [vL78]:
BAD: “In [23], algorithms are presented …”
GOOD: “Jones presents algorithms … [23].”

25. Writing Style avoid vague words (e.g., “deals with,” “handles”)
avoid contractions
be consistent in spelling, punctuation, capitalization style
use the same grammatical style for items in a list
develop flow/transitions between paragraphs, sections, chapters
avoid empty sections
merge/eliminate single item sublists or subsections
place punctuation inside quotes
avoid second person (“you …”)
try to write in only one verb tense, preferably the present tense
use “including …” instead of “etc.”
use “such as” instead of “like”
put math in definitions, theorems, proofs; explain in English to build the reader’s intuition
Use “:” instead of “-” in technical writing
Space after “)” and “:” (not before!)
Use “that” instead of “which” when not counting things

26. References
[s99] Strunk, The Elements of Style, New York: bartleby.com, 1999,
[vL78] van Leunen, M.-C., A Handbook for Scholars, Alfred A. Knopf, 1978.

27. Current Research Work Sandipto Banerjee, Ph.D.
Bartley Richardson, Ph.D.
Lydia Fitzgerald, M.S.
Bill Nicholson, M.S./Ph.D