r2 - 13 Sep 2006 - 16:54:23 GMT - VladAtanasiuYou are here: Bernstein > Main Web >  SystemsIntegration > UseCaseRequirements > UseCaseList > DbSearchHeterogeneousContent

Dabases search heterogeneous content

This is part of the submitted UseCaseList.

Scenario

A description of the scenario that you have in mind.

Although the data in the watermarks databases diverge to a certain degree in the type of content and the formats uses to represent it, the user wants means in place to handle heterogeneous content.

  1. In it's simplest form, it is required that the user-machine interaction happens through a single web portal.
  2. There should also be guides put in place drawing his attention to the heterogeneous character of the data he is about to interact with (such as time distribution of the watermarks in the cumulated databases).
  3. Responses to user queries should state what percentage of the database content the machine response refers to (for example "10.000 watermarks with laid lines density of 20 mm per 20 laid lines, from a total of 100.000 watermark images in all databases, from which 5.000 couldn't be used because the laid lines density is measured over 21 laid lines [supposing there is no software to make the conversion between the two encoding formats]").
  4. Methods should be developed to maintain as much as possible the statistical usability of the data, particularly in the case of comparison between several datasets that are defined over different populations of watermarks (For dating watermarks for example, I can combine 5 image features during expertise (WM type (name), WM proportions, WM relative portion, laid lines density, chain lines distance) knowing that their concomitant presence in a watermark description file is in much fewer numbers than the description of just one feature (the type of the WM) , which is given for most watermarks. However the two dating approaches are not similar in how the results are to be interpreted, nor do the two dates proposed by the expert system yield the same level of statistical confidence. The user needs to be guided during the dating process and the machine has to provide the necessary quantitative data about the overlapping of the used data sets.)

Importance

How important do you see this use case as?

  • Must Support

Dependencies

What other use cases are affected by the implementation of this one?

  • Databases usability / db integration / dating / authentication / cartography / bibliography

Input

Things that the user must/might supply to the system.

  • Visual search by user.

Output

Things that the user will recieve in response to their request.

  • Watermarks are found and the heterogeneity of the matching as well as the non-matching data is displayed.

Difficulties

Areas in which you foresee problems/issues arising.

  • The nature of the data provided to the user in response to his query must be made explicit, because the data is heterogeneous and because replies to multicriteria searches have different amounts of the various data types coagulated into a single reply (e.g. a survey conducted on a group where 75% of the surveyed persons are men, can't have the same significance to the same survey with 50% men).

Example

An example supporting this use case.

  • See above.

Other Information

Any other information that you think is important to include.

  • ...

Comments

Comment from other partner regarding the use case.

  • ...

-- VladAtanasiu - 13 Sep 2006

Edit | WYSIWYG | Attach | Printable | Raw View | Backlinks: Web, All Webs | History: r2 < r1 | More topic actions
 
The Bernstein Website
Copyright © by the contributing authors.
Bernstein - The Memory of Paper http://www.bernstein.oeaw.ac.at
Ideas, requests, problems regarding Bernstein? Send feedback
This site is powered by the TWiki collaboration platform