Transcript Making Database Systems Usable

Making Database Systems Usable
H.V. Jagadish Adriane Chapman Aaron Elkiss Magesh
Jayapandian Yanyao Li Arnab Nandi Cong Yu
By Shahana Shamim
• MimI A deep integration of the best-regarded protein interaction
databases
• It allows users to access data through various interfaces
• Provenance was retained to describe where the data originated
Miml
•
•
•
•
XML
XQuery
Forms-based interface
MQuery
• Complains against MimI
• Different results were found by going through different interfaces
• Inability to explore and manipulate the data directly. This issue was
addressed by integrating a graphical tool, Cytoscope which helped in
graphically manipulating the results.
• It was hard for the users to put their scientific results into MimI for easy
access by others
• Search Engine
• Directly connects users with the web
• Almost instantaneous response
• It does not address all the usability problems database systems face
Major Issues
•
•
•
•
•
•
Unexpected Pain
Unable to Query
Painful Options
Unexpected Results
Unseen Pain
Birthing Pain
Unexpected Pain
• Frustrates users when the database produces results that are unexpected
with no explanation.
• Search engine strategy does not work with database because when the
user queries the database she kind of knows what is in the database.
Whereas the web is huge and no one knows what is exactly there and
can’t point out that he or she missed something vital.
• 2 forms:
•
Unable to Query
•
Unexpected Results
Unexpected Results:
• List of Cheap Flights:
•
•
•
1) Los Angeles
$75
2) Boston
$100
3) San Francisco $400
•
Why is San Francisco in this list?
• The database should be able to explain “where” and “why”
Painful Options
* A software with too many options is costly
* As we like to limit options, it is not easy to determine which options to keep
and which options to leave out.
Unseen Pain
• A user issues a query, it doesn’t produce the required output, so she
revises the query, resubmits
• Significance of query specification is result construction
• WYSIWYG: Constant predictive capability on the part of the system
Birthing Pain
• Database evolves continuously
• Users do not have a clear idea about what the final structure of the
database will be so a comprehensive design of the database can not be
done
The database system should be able to provide interfaces for users to
create information and to fluidly manipulate the structure.
The Painless Future
• Geographic Mashups have been tremendously successful in
presenting joins between data sets using a geographic location as the
basis
• Network It is easier to point and click than to type
• Multidimensional While the data in the warehouse may be
stored in multiple tables the users think of the data as points in
multidimensional space with the aggregates of attributes.
• Tabular Excel spreadsheets show data in simple two-dimensional
tables which people like to see as joins across multiple normalized tables
may be difficult for them
Data Provenance
• What we see is most often extracted from a database, which in turn was
extracted from other databases, and so on. Provenance information has to
be understandable to the user.
Direct Data Manipulation
• Point-and-click, drag-and-drop and filling in textboxes (to a lesser extent)
improves user interaction with the database
“Schema-later” and “Heterogeneous” Database Design
**Creation of database with structured or unstructured data
** Taking advantage of the existing data structure
** Provide convenient functionality to add structure
Thank you