GEOSS Banner

Section 5: Full User stories

5.1 Supplying a comment on a problem with a dataset

User 1 wants to insert a comment on a Landsat ETM tile striping artefacts

• Using a form in the GEOSS catalogue, they create a GVQ_FeedbackItem with a primaryTarget pointing at the unique identifier of that Landsat tile, and add a subject which is a simple comment or keyword 'Landsat TM Striping problems'. They also add a secondaryTarget to the target record, which points at this whole Landsat series, to indicate that this feedback will be relevant to people interested in that secondary target.

• They add a GVQ_UsageReport to the item with a reportAspect of 'Problem'.

• They supply user information (at minimum, this will contain a CI_Role of 'User')

• In the usage report they put some text (possibly minimal) in the usageDescription to say what they did with the data.

• They also add to the usage report a GVQ_DiscoveredIssue with some text describing the knownProblem (the striping, and its nature). They can specify workarounds and alternative datasets if they like. Here, they have to give the dataset identifier for the Landsat tile again, but this can be copied from the primaryTarget information they have already supplied.

• They can add other things if they like, including information about the user's expertise, application domain and role.

Information hold in the feedback item can be searchable and thus discovered through the use of keyword on the subject or through the GVQ_DiscoveredIssue report. All these information can be very relevant to a user of that specific tile as well as to users of other Landsat ETM tiles.

5.2 Adding a secondary target

User 2 is a researcher who wants to furnish domain qualified information related to a particular dataset used on agricultural land use. The dataset is from May 2007, from the Delta del Ebre region and it contains crop codes. User 2 wants to emphasise that the particular dataset has all crop codes from the region; the user also believes to be of great interest to facilitate the links to other datasets which also use crop codes.

• User 2 creates a GVQ_FeedbackItem with a primaryTarget pointing at the unique identifier of the dataset X, and adds a subject which is a simple comment or keyword 'Agricultural - crop codes for South Wales region'.

• User 2 adds a secondaryTarget pointing at the unique identifier of a dataset Y (which also contains crop codes), to indicate that this secondary dataset may also be of interest.

• User 2 adds another secondaryTarget pointing at the unique identifier of a dataset Z that also contains crop codes but this time for May 2012. The user can add several more secondaryTargets if he/she thinks that they may be relevant.

• User 2 then adds a GVQ_UsageReport to the item which has three reportAspects: ‘FitnessForPurpose', 'Usage' and 'Alternatives'.

• User 2 supplies some user information about him/herself (at minimum, this will contain a CI_Role of 'User')

• In the usage report he/she puts some text (possibly minimal) in the usageDescription to say what he/she did with the data.

• User 2 adds a text comment to the feedback item which stresses that dataset X contains a complete set of crop codes of agricultural land use.

5.3 Adding a supplementary target

User 3 is a small producer who generates high level oceanographic datasets products from low level satellite and aircraft imagery. The User also wants to specify that one of the sensors used work better in a certain range, but there is no “hard” evidence of it. Therefore, the User wants to warn users on possible problems on the dataset produced as well as providing links to other datasets derived from that same sensor.

• User 3 creates a GVQ_FeedbackItem with a primaryTarget pointing at the unique identifier of the dataset X, and adds a subject which is a simple comment or keyword 'Sensor range issues’. A parent identifier may be automatically added to the target record, which points at this whole data series he produced. User C does not have to provide this parent information, as it should be automatically looked up within GEOSS.

• User 3 then adds a supplementaryTarget, to indicate that this feedback is also relevant to other datasets he produced using this sensor. Users who will be interested in this feedback may not directly be interested in the supplementary dataset, but providing links to one or more datasets with a similar problem could emphasise the problem with the sensor’s range, and give a useful example.

• Then he adds a GVQ_UsageReport to the item with a reportAspect of 'Problem'.

• User 3 then supplies some user information about himself (at minimum, this will contain a CI_Role of 'User') specifying that he is the producer of this data. Perhaps in this context, an obligatory rolecode and expertiseLevel would be useful.

• In the usage report he puts some text (possibly minimal) in the usageDescription to say that this dataset was not directly used by him, but he is the producer.

• User 3 also adds to the usage report a GVQ_DiscoveredIssue with some text describing the knownProblem (the problem with the sensor range).

• He can add other things if he likes, including information about his expertise, application domain and role (producer).

5.4 Citing a publication which uses specific data

User 4 wants to inform of a publication that uses MODIS data for cropland mapping for a time period. This feedback is of interest to users looking for publications on MODIS data or to cropland mapping.

• She creates a GVQ_FeedbackItem with a primaryTarget pointing at the identifier of the global product, and adds to the target a GVQ_Focus which is datacentric - it has a temporal extent in it which describes the temporal footprint of the data she used. She also adds a domainURN to the GVQ_FeedbackItem which reports (by mapping to a GEMET or other concept) that she focussed on cropland mapping.

• She supplies some user information about herself (at minimum, this will contain a CI_Role of 'User').

• She adds a GVQ_UsageReport to the item whose reportAspect is 'Usage'.

• In the usage report she puts an abstract in the usageDescription to say what she did with the data.

• She adds a citation to the GVQ_FeedbackItem, with the publication information in it.

• She can add other things if she likes, as above. A client that allows users to report publications could force them to add some contextual information.

Someone looking for information on MODIS data or on cropland mapping would be interested on this feedback reported.

The same issues of search by identifier or keyword arise as for the case above, and are common to practically all useful searches. We haven't explicitly modelled a database of identifiers and their parent-child relationships in the feedback model because we can't promise or hope to maintain this in GeoViQua. What we have done is add many opportunities for sensible identifiers to be used which can later be resolved by software clients.

5.5 Reporting intercomparison of datasets, and fitness-for-purpose assessment

User institution 5 wants to report some serious frequency spike artefacts found when validating the updated contour map generated for a small area  from a very high quality DEM and compares it to the national cadastral product. The metadata conforms to the producer quality model.

• They create a GVQ_FeedbackItem with a primaryTarget pointing at the national contour product, and add to that target a datacentric focus containing a spatial extent which describes the spatial footprint of the area they considered. They may add other descriptive text or domainURN information about the field in which they are working.

• They supply some user information about themself (at minimum, this will contain a CI_Role of 'User'). It can contain plenty of other information, and perhaps in this context, an obligatory userRole and expertiseLevel would be useful.

• They add a GVQ_UsageReport to the item which has three reportAspects: 'FitnessForPurpose', 'Alternatives' and 'Problem'.

• In the GVQ_UsageReport they put some information in the usageDescription to say what they did with the data.

• They also add to the usage report a DiscoveredIssue with some text describing the knownProblem (the frequency spikes). As an alternative dataset they cite their new contour data.

• They add to the feedback item a GVQ_QualityOverride with fuller numerical and statistical information on the issue they discovered. It's a GVQ_DataQuality, so it has a reference to their data as the MD_AssociatedResource in the context of crossReference, and a reference to a fuller publication on the subject.

This kind of feedback is of interest to people who wants to work with alternatives to the national contour products. Some search on the quality of the data (through publications that may refer to the dataset, reported evaluation, figures) as well as looking at the user information to determine if it is qualified could also be necessary.

5.6 Constructing a ‘GEOLabel’ for a dataset

User 6 wants to gather all information related to the GEOLaber (concept described below) for a particular dataset. This can be done through queries at several levels which are sent to the producer and to the user metadata.

Level 1 (basic information)

Some queries are on the existence of quality statements for a particular dataset, on quality statement from the producer for a parent dataset, on the availability of user feedback or on information on the average user rating.

Level 2 (further details that can be looked into

Some queries on Level 2 are related to the quality statement, on how complete are the metadata , on showing a summary on what the user feedback contains, on what are the individual user ratings and associated comments,...

5.7 Providing general feedback on a metadata record

User 8 is a climate forecaster who uses climate raster data X from a website which doesn’t provide much information on the available datasets. User 8 has been running some analysis on it, and has found that the the data is of good quality and that the metadata complies with ISO 19139 standard, but relevant information to him on lineage is not present. To assess the degree of population of the metadata he/she uses the online NOAA/GeoViQUa Rurbic (Shown in figure 31), wants to report it to potential users, and  add some comments on the issue he found disappointing.

• User 8 creates a GVQ_FeedbackItem with a primaryTarget pointing at the unique identifier of that metadata record, and adds a subject which is a simple comment or keyword 'Metadata record completeness problems'.

• User 8 adds a GVQ_UsageReport to the item with a reportAspect of 'Problem'.

• He supplies some user information about himself (at minimum, this will contain a CI_Role of 'User').

• In the usage report he puts some text (possibly minimal) in the usageDescription to say how he assessed the data and metadata record.

• User 8 also adds to the GVQ_UsageReport a GVQ_DiscoveredIssue with some text describing the knownProblem (incomplete metadata record, no conformance to international standards).

• He adds a publication referenceDoc which contains an onlineResource pointing at the NOAA / GeoViqua rubric analysis of this metadata document (see Figure 31, next page).

An XML encoding of this feedback element can be seen at:

http://uncertgeo.aston.ac.uk/GVQ_v2_...7_metadata.xml.

Rubric.png

Figure 31. A screen shot of the NOAA / GeoViQua completeness evaluation rubric for metadata

      

5.8 Providing general feedback on a specific metadata record elements

User 9 an archivist on land use data who has found an aerial photography dataset X on data land use in Girona in 1946. Some analysis were performed on both metadata and data, and it was found that the spatial coverage of the photography didnt correspond to the one documented in the metadata. User 9 wants to report this issue with the correct metadata coordinates, and give some additional information on other datasets from Girona region.

• User 9 creates a GVQ_FeedbackItem with a primaryTarget pointing at the unique identifier of the metadata element, and adds a subject which is a simple comment or keyword 'Metadata spatial coverage attribute problem'.

• User 9 also adds a secondaryTarget pointing at the unique identifier of other dataset that was used as a comparison.

• She adds a GVQ_UsageReport to the item with two reportAspects: 'Problem' and 'Alternative'.

• She supplies some user information about herself (at minimum, this will contain a CI_Role of 'User'). In this context, an obligatory rolecode and expertise level would be useful.

• In the usage report she puts some text (possibly minimal) in the usageDescription to say how data and metadata record were analysed.

User 9 also adds to the usagereport a GVQ_DiscoveredIssue with some text describing the knownProblem (spatial coverage not being correct). The user can specify workarounds (provide actual spatial coverage information), and alternativeDatasets.

                

5.9 Searching for a domain-specific rating on a resource

User 10 wants the average dataset rating resulting from usage by hydrologists FOR A DATSET, AND FOR THE SERIES TO WHICH THAT DATASET BELONGS. How to do it: Find the identifier of the parent dataset, and for both identifiers, find feedback records whose GVQ_ReportAspectCode is 'Usage' and whose GVQ_UserInformation>applicationDomain or GVQ_FeedbackItem>domainURN is a urn mapping to 'hydrology'. Then aggregate any ratings in those records.

5.10 Searching for information on problems with a dataset for a specific space/time window.

User 11 is interested on gathering any problems reported by research end-users who have been working with datasets in Australia between 1987 and 1995.

How to do it: Find records where this data is the primary target. In the case of the feedback server, query the GVQ_UserRoleCode to find research end-users and find coincident extents with the spatial and temporal extent of any datacentric focus elements. Look for discoveredIssue elements. If required, the search can be further refined by looking for usageReports with a reportAspect of ‘Problem’.

5.11 Searching for individual products which are cited in more than 15 unique journal articles

How to do it: Searching GVQ_Publication records associated with GVQ_FeedbackItems. A simple search would simply count the number of unique citations (relying on ISSN) per resource. A more sophisticated search might take the hierarchy of data resources into account, and include publications relating to subsets of the specified resource. This would require access to an external taxonomy of GEOSS resources and their relationships.

5.12 Finding datasets which meet user-specified quality thresholds

Finding datasets that meet the quality thresholds  can be done interrogating the metadata documents registered in GEOSS by a producer. Some examples on the searches that can be done with the current standards are given (example 1 and 2), it is also given an example of a more complex search that will be done when the GeoViQua model is implemented.

1. Find all datasets with a RMSE Positional Accuracy better than 45m.

Great care has to be taken when doing this search since there could be several issues related to the units and measure, some datasets may not document one of these fields and thus the search could be wrong.

2. Find me datasets with a pixel resolution smaller than 400m, and a thematic classification correctness greater than 75%

3. Find me datasets whose quality was assessed with reference to UK Ordnance Survey's Mastermap.

How to do it: Get the dataset identifier(s) for the Mastermap product and look for it/them in the referenceDataset of any reported quality evaluation elements.

4. Find me datasets with an attribute accuracy < 1degC^2 for all pixels / locations in my region of interest.

How to do it: This is a very tricky example, but if we had pixel / object level information on accuracy, it can potentially be carried out. Whether this can be done within the scope of GeoViQua remains to be seen!


<<Prev T.O.C. 
Tag
none

Files (0)

 
You must login to post a comment.