Observation on Database Research Trends via Publication Statistics

1
Observationon Database Research Trendsvia
Publication Statistics

• Amanuel Godefa
• David Kuo-Wei Hsu

2
Outline

• Problem Statement
• Contribution
• Our Approach
• Results
• Conclusion
• Future Work
• Assumptions

3
Problem Statement

• Given
• Database research papers - VLDB SIGMOD,
  19972006
• 1170 regular papers (22 groups)
• Find
• what will be the next research topics in the
  future, by observing on the database research
  trends via publication statistics .
• Objective
• Provide a statistical analysis and conclusion on
  popular topics of DB research paper (now near
  future)
• Constraint
• Collect data manually
• Statistical survey

4
Contributions

• Statistical survey of publications in two major
  conferences, (VLDB and SIGMOD)
• compare these two significant conferences in
  terms of topics or research areas they prefer
• Analysis about future research topics in database
  area
• New researchers can use this statistical survey
  as a reference to choose (or change) research
  topics.
• Lecturers can design new database courses based
  on the solid support provided by this statistical
  survey.

5
Our Approach

• VLDB and SIGMOD conferences 1997 to 2006
• only on the regular or research session
• read abstract of 1170 papers
• 1579 topics counted
• categorized into 22 groups

6
Results on comparison on SIGMOD VLDB
Publications
7
Results Combined Data
8
Results
9
Results
10
Results
11
Summary for Five Year Interval Only for Those
With Higher Percentile Differences, 35 and Above
or 35 and Below
12
Conclusion

• Smile curve
• Stop broadening
• Start deepening
• Good-to-go topics (Can you Guess?)
• Query processing (evergreen always on the top,
  about 28 over all)
• Stream data management
• Sensor network
• Security/privacy
• Distributed environment

13
Future Works

• Statistical survey for other conferences and
  journals (or even for longer sampling periods),
• Developing tools (e.g., crawler)
• Consider the industries and the development of
  commercial products

14
Assumptions

• We use a 0.05 for all T-Test
• Both sampled populations have distributions that
  are approximately normal
• For detailed statistical test, please visit
• http//www-users.cs.umn.edu/godefa/CSci8701/P10_F
  inal.doc

15

• Wishing You the Best in Your Finals
• Happy Holidays