Visualizing Data How Content-Based Queries and Graphical Representation Will Change the Way we Manage Data Frank, M., “Fut...
Emerging Trends <ul><li>Content-based queries for multimedia data </li></ul><ul><ul><li>surpass the limitations of keyword...
Querying Multimedia Data <ul><li>Querying multimedia data has been an indirect process </li></ul><ul><li>Users had to rely...
Potential Applications <ul><li>Online shopping catalogs </li></ul><ul><li>Advertising agencies </li></ul><ul><li>Stock pro...
Vendors of Content-Based Queries Software <ul><li>IBM: Ultimedia Manager for OS/2 </li></ul><ul><ul><li>Query By Image Con...
Ultimedia Manager <ul><li>Runs on OS/2, AIX and Windows </li></ul><ul><li>Works in conjunction with DB2 </li></ul><ul><li>...
QBIC Technology <ul><li>Five image features: </li></ul><ul><ul><li>color (two measures) </li></ul></ul><ul><ul><li>texture...
“Computing the image” <ul><ul><li>color </li></ul></ul><ul><ul><ul><li>adding up the red, green, an dblue pixels </li></ul...
“Computing the image” <ul><ul><li>texture is analyzed in three ways </li></ul></ul><ul><ul><ul><li>coarsenes </li></ul></u...
Querying and Browsing <ul><li>Users can create visual queries </li></ul><ul><ul><li>alphanumeric search criteria defined w...
Examples
 
 
 
Data Visualization <ul><li>Visual display of information to support recognition of patterns and anomalies by an analyst </...
Visual Data Mining <ul><li>Intermediate data storage layer that precalcualtes, indexes, and caches summarized results need...
Target Markets <ul><li>Analysis-intensive companies in the financial sector </li></ul><ul><li>Brand-oriented companies suc...
 
Upcoming SlideShare
Loading in …5
×

Data Visualization

451 views
360 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
451
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
6
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Data Visualization

  1. 1. Visualizing Data How Content-Based Queries and Graphical Representation Will Change the Way we Manage Data Frank, M., “Future Database Technologies,” DBMS , 8, 12 (Nov. 1995), pp. 52-58. http://wwwqbic.almaden.ibm.com/%7eqbic/qbic.html http://www.research.microsoft.com/research/datamine/ IEEE Spectrum, Nov. 1995
  2. 2. Emerging Trends <ul><li>Content-based queries for multimedia data </li></ul><ul><ul><li>surpass the limitations of keyword searches for graphics </li></ul></ul><ul><li>New 3D visulatization techniques </li></ul><ul><ul><li>users surf through graphical representations of data at various levels of detail </li></ul></ul><ul><li>Both depend on powerful computers </li></ul>
  3. 3. Querying Multimedia Data <ul><li>Querying multimedia data has been an indirect process </li></ul><ul><li>Users had to rely on associated keywords, primary key values, or other alphanumeric fields </li></ul><ul><ul><li>different people describe images, sounds, video in different ways </li></ul></ul><ul><li>Managing multimedia resources as a database </li></ul><ul><ul><li>not just a collection of files in a directory </li></ul></ul><ul><ul><li>must be examined before they are added to a database </li></ul></ul><ul><ul><li>stored as BLOBs or as operating system files with their location traced by the database </li></ul></ul><ul><ul><li>a variety of numeric measures of the multimedia objects’ physical characteristics are derived </li></ul></ul>
  4. 4. Potential Applications <ul><li>Online shopping catalogs </li></ul><ul><li>Advertising agencies </li></ul><ul><li>Stock protography service bureaus </li></ul><ul><li>Medical services (X-rays) </li></ul><ul><li>Trademark searches </li></ul><ul><li>Film producers (including training departments and training companies) </li></ul><ul><li>Electronic publishers </li></ul><ul><li>Satellite and weather photograph analysis </li></ul><ul><li>Military applications </li></ul>
  5. 5. Vendors of Content-Based Queries Software <ul><li>IBM: Ultimedia Manager for OS/2 </li></ul><ul><ul><li>Query By Image Content (QBIC - pronounced “cubic”) </li></ul></ul><ul><ul><li>“ Relational Extenders” for DB2 promises to upsize this capability </li></ul></ul><ul><li>Virage Inc. offers similar capabilities to developers using any DBMS </li></ul><ul><li>Silicon Graphics Inc. hopes to enliven the data warehouses with visual query metaphors </li></ul>
  6. 6. Ultimedia Manager <ul><li>Runs on OS/2, AIX and Windows </li></ul><ul><li>Works in conjunction with DB2 </li></ul><ul><li>Is a front-end application </li></ul><ul><ul><li>provides GUI for querying multimedia objects based on: </li></ul></ul><ul><ul><ul><li>content, </li></ul></ul></ul><ul><ul><ul><li>alphanumeric descriptive fields, or </li></ul></ul></ul><ul><ul><ul><li>combination of both </li></ul></ul></ul><ul><ul><li>lets users insert multimedia objects into a DB2 database </li></ul></ul><ul><ul><li>produces a catalog of objects </li></ul></ul><ul><ul><ul><li>resides in a DB2 database </li></ul></ul></ul><ul><ul><ul><li>stores mathematical representations of each multimedia objet’s physical characteristics </li></ul></ul></ul>
  7. 7. QBIC Technology <ul><li>Five image features: </li></ul><ul><ul><li>color (two measures) </li></ul></ul><ul><ul><li>texture </li></ul></ul><ul><ul><li>shape </li></ul></ul><ul><ul><li>position </li></ul></ul><ul><li>The numerical scores become indexes into an image database </li></ul><ul><li>Can analyze the entire image or one or more defined regions within the image </li></ul>
  8. 8. “Computing the image” <ul><ul><li>color </li></ul></ul><ul><ul><ul><li>adding up the red, green, an dblue pixels </li></ul></ul></ul><ul><ul><ul><ul><li>works with small number of colors </li></ul></ul></ul></ul><ul><ul><ul><li>a color histogram </li></ul></ul></ul><ul><ul><ul><ul><li>measures the distribution of colors into 64 color ranges </li></ul></ul></ul></ul><ul><ul><li>shape </li></ul></ul><ul><ul><ul><li>depends on up to 18 geometric variables such as circularity or the orientation of a major axis </li></ul></ul></ul>
  9. 9. “Computing the image” <ul><ul><li>texture is analyzed in three ways </li></ul></ul><ul><ul><ul><li>coarsenes </li></ul></ul></ul><ul><ul><ul><ul><li>the average size of areas with similar intensity </li></ul></ul></ul></ul><ul><ul><ul><li>contrast </li></ul></ul></ul><ul><ul><ul><ul><li>the variation between dark and light areas </li></ul></ul></ul></ul><ul><ul><ul><li>directionality </li></ul></ul></ul><ul><ul><ul><ul><li>indicates whether edges are more or less regularly spaced (grass) or randomly distributed (leaves) </li></ul></ul></ul></ul><ul><ul><li>position </li></ul></ul><ul><ul><ul><li>represents an area’s location within an image </li></ul></ul></ul>
  10. 10. Querying and Browsing <ul><li>Users can create visual queries </li></ul><ul><ul><li>alphanumeric search criteria defined with graphical dialog </li></ul></ul><ul><ul><li>visual queries </li></ul></ul><ul><ul><ul><li>shape can be drawn or selected form a collection of sample shapes </li></ul></ul></ul><ul><ul><ul><li>a color wheel with slider controls to adjust colors </li></ul></ul></ul><ul><ul><ul><li>texture is selected from a palette of sample textures </li></ul></ul></ul><ul><ul><ul><li>the location of the sample image indicates the position </li></ul></ul></ul><ul><li>Query results are displayed as: </li></ul><ul><ul><li>two thumbnails (whole picture and a matching detail) </li></ul></ul><ul><ul><li>alphanumeric data, if used in the query </li></ul></ul><ul><ul><li>images are sorted based on their computed similarity </li></ul></ul><ul><ul><li>queries can be refined by changing weights </li></ul></ul><ul><ul><li>an image from the output can be used as new criteria </li></ul></ul>
  11. 11. Examples
  12. 15. Data Visualization <ul><li>Visual display of information to support recognition of patterns and anomalies by an analyst </li></ul><ul><li>Visual querying of old-fachioned transactional data </li></ul><ul><li>Data mining of large data warehouses </li></ul><ul><li>An alternative 3D image image-oriented approach to drill down and data exploration </li></ul><ul><li>User begins with images like 3D maps or graphics representing data groupings such as product categories, corporate divisions, or demographic groups </li></ul><ul><li>Users click on these to access the underlying data </li></ul>
  13. 16. Visual Data Mining <ul><li>Intermediate data storage layer that precalcualtes, indexes, and caches summarized results needed for drill down </li></ul><ul><li>Similar to OLAP </li></ul><ul><li>SGI sees data visualization and OLAP engines sharing the same intermediate layer within a data warehouse environment </li></ul><ul><li>Targeted at executives and managers unwilling to learn complex query and analysis programs </li></ul><ul><li>SGI’s introductory product will have an API that C or C++ programmers can use </li></ul>
  14. 17. Target Markets <ul><li>Analysis-intensive companies in the financial sector </li></ul><ul><li>Brand-oriented companies such as consumer goods manufacturers </li></ul><ul><li>SGI has begun a limited beta testiing </li></ul><ul><li>First shipments in early 1996 </li></ul><ul><li>Current offerings are more akin to an appetizer than an entree </li></ul><ul><li>The best is yet to come </li></ul>

×