NOAA Data Management Dashboard

NOAA – Data Management Dashboard (Matt Austin)

• This is still a work in progress
• Visual tool for NOAA data management standards
• We all need to work together
• Need to review mark-up before do any design
• Can show number of records that satisfy Documentation directive
  • Sum of all data management plans
  • Need to look at how much it is being used
  • These metrics can be shown on home page
  • By having observing system – would save time in office & have better results
  • Have to define what a dataset is
  • Can have drill down options – look at how metrics change with training
• Need a consistent way to implement standards – like what Documentation cluster is looking at
  • Find ways to aggregate metadata to answer questions
• What high level design makers need – prioritize questions
• Sources – can use web accessible
• This can improve interoperability of NOAA data
• Can organize dashboard at multiple levels – end user usually wants things topically
  • Have not fully defined collection vs. grandul
• http://sites.google.com/a/noaa.gov/noaa-hppc-shared-hosting-project/  (need permissions)
  • NOAA has switched to google
  • Number of metadata records in NOAA Geopotal
  • See changes in number of records – number of records does not mean always better
  • Google spread sheet can update in real time – or can use a database
• Q – do you have to use geoportal – no but NOAA is
  • Need to focus on standards so that it is the same for everyone
• Q – are these available anywhere else
  • Can share them (will get screen shots)
• Q this is just a concept
  • Google doc – shows what development that is finished
• Ted has more

Dashboard Displays for Classifying Metadata Collections (Ted Habermann)

• A dashboard in your car – metrics about the system
  • Without an engine – dashboard is not interesting
  • Interested in how to build in NOAA, communities/engines worth reporting on
  • One target audiences – CIO  or management – if nothing changes on dashboard – it is a problem
  • How can we get the community to participate
• Work on set of data sets – related to project or other – rows in table are collection of records from NOAA
  • Uses a rubric to measure completeness (0-41) – provide: average score, std. dev. Total, counts above 20 and 25
  • When started with rubric – thought it was for single records – but found interesting information
  • One set has the same min, max, and average and some have small std dev
    • Includes link to histogram of score – see different characteristics between data groups
  • IOOS – metadata generated automatically
    • Have more at 45
    • Q – count seem low – these are specifically related to sensor operation records or THREDD records
  • Mean and std dev – recognize records with different characteristics
  • If look at table of contents for records – can look at individual record rubric – which is navigable and can see the fields and links to the wiki to improve the metadata
    • At bottom – other reports – ex. Broken uls, broken xlinks, validation errors, unique contacts
    • Created contact reports – to see who to contact if have problems
    • Looking at common content/aggregation is a powerful too
  • Consistency checker
    • Takes all of the records (usually around 100) – aggregate into single xml file – then look for repeated content
    • Connections, people, citations, services (can sort by fields in each tab) – provides a count for each
    • This provides a means to look for inconsistencies that occur in record
    • Q – (Matt) – seems like the kind of thing that can be automated to not burden the people that are busy need to change
      • Not sure of tool that will be able to figure out a typo/formatting
      • Ted would make change in oxygen (grouped metadata) – find and reply then it will save it back to the original
      • (Matt) – if you provide to the researcher – may not care about problem
    • Dashboard is a tool to help people creating the metadata
  •  Q – as a search engine – if doing textual then some of these may not come up because of typos
    • exactly
  • sometimes need to look at code for fix to the problem – others are human errors that need to be looked at – machine generated vs hand hewn records
  • if change from rank to time – see same variability
    • to the manager – the older records are finished – so translation from FGDC to ISO
      • for older records – want to translate without major problems (Pandora’s box) – if it ant broke-don’t fix it
      • with newer records – want the data manger to work with the ISO standards
      • helps develop strategies for moving forward – strategic tool for guiding the direction and indentifying the approaches to help evolve the organization
• Q Eric (Kodak) – are you indenting to define dashboard for end user – or widgets that any user can customize
  • Customize is good – want to start with “My” – for a specific collect defined by string or set of options
  • Not meant to change settings – it sit and changes results
• Q – (Eric) like that you can relate to dashboard – use can depend on user and context
  • Makes it better if users can customize
  • Identify what users are interested
  • Then you can query to see who is using what
  • In junked car – still fuel and odometers (fuel = explosion)… some are universal – might be really hard for all audiences
• ACTION - If there are tools that might be useful – please let Ted or Matt know
  • EXTJS
• Really have to think about data model – the limitations of the metadata – if can build data model to think about potential uses – right now for data stewards – but can think of how other use it
• Q (Don Collins – NODC) – wants to see what NOAA is asking about – Matt described what might get measured, quantified, reported – Ted focused was to quantify/target to improve data for NOAA – wants to know where metadata is good or not/need improvements OR is this some measurement to see if system is working – have not solidified internal concept
  • Remember Bruce Carin’s talk -… this is a coffee house
• Ted testing – Matt is see what documentation needs to be done
  • Less about – what question for who MORE how can you answer  any question for any reason
  • Reason for doing this in ESIP – because want to share outside of NOAA – reach mass market audience
• Heather Hinkle (USGS) – survey wide data management program (starting) – metrics will happen in ~ phase 2 – show growth/standards – may want sooner – metric are tricky – need to be answering the right question – show both quality and quantity
• Want to start metrics early – usually will create metadata and then “later” improve it
• Ted – also try bring to data.gov – looking for data stewards
• (Erik) – difference between metric and the interpretation – give number, but don’t interpret – give guidelines to interpret
  • Challenge #1 – exposing metadata
  • Challenge #2 – finding the problems
  • Challenge #3 – fix the problems in the metadata (people don’t want to fix)
  • Ted – other thing is building relationship with people that are doing these things