Sample Use Cases for the FUSION tools

This document contains a number of example queries and discussion of the information required to evaluate and respond to them. There are a number of alternatives algorithms for actually applying and structuring the querying system which depend on the way in which the available data is stored. This document is not intended to describe or support any particular methods but merely to investigate, informally, the process, semantic connections and information required to fulfill some of the purposes of the FUSION system.

Each query is presented in plain english (although it is unlikely the final system will provide this capability) with terms from the ontologies underlined and values in italics. A possible proceedure for finding the required information is presented, followed by any example xml and then each step of the proceedure is discussed in point form.

Use Case 1

Use Case 2

Use Case 3

Use Case 1

How does the mean catalyst size in electrodes of width < 20 microns effect electrode conductivity?

Need to:

Find all electrodes of width less than 20 microns
Find performance/conductivity data for those fuel cells
Find the mean catalyst size for those electrodes

image.xml: Contains information such as shape, size, colour, density, texture, location of each feature/segment of the image as found by the image analysis software (or web service) and transformed during the image analysis phase to fit the image analysis schema.

<?xml version="1.0" encoding="iso-8859-1"?>
<Mpeg7 xmlns="urn:mpeg:mpeg7:schema:2001" 
          xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
          xmlns:mpeg7="urn:mpeg:mpeg7:schema:2001" 
          xmlns:xml="http://www.w3.org/XML/1998/namespace" 
          xsi:schemaLocation="urn:mpeg:mpeg7:schema:2001 .\Mpeg7-2001.xsd">

  <Description xsi:type="ContentEntityType">
    <MultimediaContent xsi:type="ImageType">
      <Image>
        <MediaLocator>
           <MediaUri>http://metadata.net/FUSION/media/image.jp2</MediaUri>
        </MediaLocator>
        <CreationInformation>
           <CreationUri>http://metadata.net/FUSION/microscopy.xml</CreationUri>
        </CreationInformation>
        <MediaInformation>
           <MediaProfile>
              <MediaFormat>
                <Content href="urn:mpeg:mpeg7:cs:ContentCS:2001:2">
                     <Name>visual</Name>
                </Content>
                <FileFormat href="urn:mpeg:mpeg7:cs:FileFormatCS:2001:3">
                   <Name xml:lang="en">JPEG2000</Name>
                </FileFormat>
                <FileSize>10483</FileSize>
                <VisualCoding>
                   <Format href="urn:mpeg:mpeg7:cs:VisualCodingFormatCS:2001:1" colorDomain="binary">
                     <Name xml:lang="en">JPEG2000</Name>
                   </Format>
                   <Pixel aspectRatio="0.75" bitsPer="8"/>
                   <Frame height="288" width="352" rate="25"/>
                </VisualCoding>
              </MediaFormat>
            </MediaProfile>
        </MediaInformation>
        <SpatialDecomposition>
           <StillImage>
              <Shape type="RegionShape">circle</Shape>
              <Color type="DominantColor">(118,118,118)</Color>
              <Texture type="EdgeHistogram"></Texture>
           </StillImage>
        </SpatialDecomposition>
      </Image>
    </MultimediaContent>
  </Description>
</Mpeg7>

rules.xml: Contains rules which link properties of image features to semantic terms in the FUSION ontology.

<?xml version="1.0" encoding="UTF-8"?>
<rulebase xmlns="http://www.ruleml.org/" 
          xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
          xmlns:mpeg7="urn:mpeg:mpeg7:schema:2001"
          xmlns:mathml="http://www.w3.org/1998/Math/MathML" 
          xmlns:xml="http://www.w3.org/XML/1998/namespace"
          xmlns:fusion="http://metadata.net/FUSION">
   <imp> 
<!-- A region depicts Electrode if the dominant colour value is 
        (118, 118, 118) and the shape is logically equivalent to a circle 
        and the texture is . -->
      <_head>
         <atom>
            <_opr>
               <rel>mpeg7:depicts</rel>
            </_opr>
            <var>Region</var>
            <ind>fusion:Electrode</ind>
         </atom>
      </_head>
      <_body>
         <and>
            <atom>
               <_opr>
                  <rel>mpeg7:DominantColor</rel>
               </_opr>
               <var>Region</var>
               <ind>(118,118,118)</ind>
            </atom>
            <atom>
               <_opr>
                  <rel>mpeg7:Shape</rel>
               </_opr>
               <var>Region</var>
               <ind>circle</ind>
            </atom>
            <atom>
               <_opr>
                  <rel>mpeg7:Texture</rel>
               </_opr>
               <var>Region</var>
               <ind></ind>
            </atom>
         </and>
      </_body>
   </imp>
</rulebase>

1. Find all electrodes with width < 20 microns

The image regions which represent electrodes are determined by applying inferencing rules to the fuel cell images
The width is calculated from the feature.length property value for image regions that are electrodes - this may be difficult due to different magnifications of the microcopy images and the need for conversions between units e.g., 16 500 nm = 16.5 microns which is < 20 microns and fits the requirements
The image.xml files are retrieved for
The image.xml has a link to the microscopy.xml which contains the id of the fuel cell.

2. Find performance/conductivity data for those fuel cells

The Fuel Cell ID is used to access the performance.xml data which records the AC impedance test results which describe electrode conductivity

3. Find the mean catalyst size for those electrodes

Using the Fuel Cell ID, all related images can be retrieved
Using the rule which describes which features are "catalyst", their area (which is a subclass of size, this is known from the ontology) can be derived
Using the MathML encoded rule for calculating mean, the mean value for all of the returned catalyst sizes for each electrode can be calculated

Finally present all results to the user, get preferred display relationships and generate presentation.

Use Case 2

Show me cathodes manufactured by 'Ruth' with thickness greater than 25 microns and high catalyst density.

Need to:

Get the id of all fuel cells (cathodes) manufactured by 'Ruth'
Retrieve relevant images by applying the cathode inference rule
Filter the results to find those images which show cathodes with thickness > 25 microns and high catalyst density by applying rules to determine the thickness and catalyst density (or since the db should already be populated by the inferencing engine just apply the appropriate sql)

1. Get the id of all fuel cells (cathodes) manufactured by 'Ruth'

this information is contained in the manufacturing.xml document
however if we were to have a relational db - SELECT fuelcell_id FROM manufacturing WHERE manufacturer IS 'Ruth';
from the fusion ontology we know that a cathode is a component of a fuel cell

2. Retrieve relevant images by applying the cathode inference rule

we have the fuel cell id so we can retrieve all the images of this fuel cell
by looking at the inferencing rule for determining if the image displays the cathode we can select those images which show the cathode of the fuel cells selected

3. Filter the results to find those images which show cathodes with thickness > 25 microns and high catalyst density by applying rules to determine the thickness and catalyst density (or since the db should already be populated by the inferencing engine just apply the appropriate sql)

theoretically the inferencing of the thickness (or width) and density information could be done on the fly but in reality it is more likely that this data will have already been inferred and stored for easy retrieval
if this is the case then we only need to retrieve the data about the fuel cells already selected and select those which meet our criteria
one interesting point in this query is defining the meaning of 'high'. A number of options exist including requesting clarification from the user in the form of a lower bound value; basing the selection criteria on past information from this user or other users; defining quantifiers for terms such as 'high' in external rules for each aspect (density, thickness etc)

Use Case 3

Show me the AC impediance values over time for fuel cells with efficiency greater than 75% and cost lower than 35 USD/hr.

Need to:

Find all fuel cells (ID) where the efficiency is > 75% and the cost is < 35 USD/hr
Retrieve the AC impediance measurements for the select fuel cells
Use the gathered data to generate a graph showing the AC impediance values over time (don't even know if that makes sense!)

1. Find all fuel cells (ID) where the efficiency is > 75% and the cost is < 35 USD/hr

data is in performance.xml
if the data is in a db then it's a simple SQL query: SELECT fuelcell_id FROM performance WHERE efficiency > 75 AND cost < 35;
as mentioned in the first use case there is also the issue of converting between different units to find equivalent values

2. Retrieve the AC impediance measurements for the select fuel cells

again a relatively simple query regardless of the storage method

3. Use the gathered data to generate a graph showing the AC impediance values over time (don't even know if that makes sense!)

possible display methods include using SVG to generate a visual graphic or just a simple html style table of values

Once the set of relevant fuel cell ids have been found based on whatever manufacturing, performance or physical attribute restrictions the user has choosen, the information that can be displayed is essentially any data or media which is indexed by fuel cell id. The links between information related to fuel cells is described in other documents. As a generalisation it could be said that the procedure here is to find the set of fuel cells (the fuel cell id which is 'universal' within the FUSION system) that are of interest and then offer the user the opportunity to select the information they want to see and how they want it displayed (re: Katya's work). For example the user could select fuel cells which meet certain performance related criteria (as in use case 3) but be interested in the values of physical attributes such as porosity and thickness and therefore request that the images of these fuel cells showing certain components together with graphs depicting the thickness compared to cost be displayed.