Difference between revisions of "Community concept mapping activity March 2020 through July 2020"

From Earth Science Information Partners (ESIP)
Line 10: Line 10:
 
<BR>
 
<BR>
 
<BR>
 
<BR>
'''A note to those who tend to lose sleep over knowledge representation and hopes of eventually watching a crossover movie titled ''Terminator vs. WestWorld Hosts vs. Blade Runner replicants vs. Wall-E trash robots'':''' For the foreseeable future, the c-soup is almost certainly '''NOT''' machine-readable.  It is true that the concept maps can be easily exported to an XML formatted file for parsing, but the nodes and relations (i.e. subject-predicate-object triples) contain text that are '''NOT''' well-formed expressions.  These ill-formed expressions run the gamut of conceptually tight expressions that correspond to terms in a given ontology (e.g. relations labeled as "has part") to human-readable but machine-goes-bonkus-because-of-badly-formed-expressions (e.g. "Decisionmaking use cases built on science, technology, policy components"... machine goes 'huh'?).  Some nodes in the c-soup are semantically annotated on an experimental basis.  Individual concepts may ultimately (not done in this experiment) be represented as subject-predicate-object triples.  Data-to-decisions traceability maps may be constructed by ordering the triples.  These triples and ordered-triple-sequences may then further be exposed to algorithms for machine-assisted discovery.   
+
'''A note to those who tend to lose sleep over knowledge representation and hopes of eventually watching a crossover movie titled ''Terminator vs. Westworld hosts vs. Blade Runner replicants vs. Wall-E trash robots'':''' For the foreseeable future, the c-soup is almost certainly '''NOT''' machine-readable.  It is true that the concept maps can be easily exported to an XML formatted file for parsing, but the nodes and relations (i.e. subject-predicate-object triples) contain text that are '''NOT''' well-formed expressions.  These ill-formed expressions run the gamut of conceptually tight expressions that correspond to terms in a given ontology (e.g. relations labeled as "has part") to human-readable but machine-goes-bonkus-because-of-badly-formed-expressions (e.g. "Decisionmaking use cases built on science, technology, policy components"... machine goes 'huh'?).  Some nodes in the c-soup are semantically annotated on an experimental basis.  Individual concepts may ultimately (not done in this experiment) be represented as subject-predicate-object triples.  Data-to-decisions traceability maps may be constructed by ordering the triples.  These triples and ordered-triple-sequences may then further be exposed to algorithms for machine-assisted discovery.   
 
<BR>
 
<BR>
 
<BR>
 
<BR>

Revision as of 05:47, April 6, 2020

Objective of this experimental concept map repository

Between March 2020 and July 2020, we aim to create a small collection of concept maps ("c-soup") that ESIP members are encouraged to contribute to. You could also use existing concepts in the c-soup to assemble a high-level concept-map that is really a narrative about how some technology you're working on (regardless of its technology readiness level) relates to an applied problem. That high-level concept-map may be thought of as a "data to decisions" pathway, or a traceability map of sorts. Program managers can use these pathways as a narrative to rapidly assess the conceptual landscape that contextualizes your technology and its potential impact.

The schema below represents one of many possible ways to conceptualize the c-soup.

For the time being, the concept maps are meant more as a communication and learning tool for humans rather than for machines to reason with. The c-soup is therefore meant to foment discussions about what it will take to: (1) use concept maps as stakeholder communication tools for applications like climate resilience planning, (2) re-use the concept maps for machine-assisted discovery of relevant digital research objects to aid in science-informed, data-driven decision making. See the section below titled "Why we're doing this" for a slightly extended treatment of the rationale for this experimental prototype.

A note to those who tend to lose sleep over knowledge representation and hopes of eventually watching a crossover movie titled Terminator vs. Westworld hosts vs. Blade Runner replicants vs. Wall-E trash robots: For the foreseeable future, the c-soup is almost certainly NOT machine-readable. It is true that the concept maps can be easily exported to an XML formatted file for parsing, but the nodes and relations (i.e. subject-predicate-object triples) contain text that are NOT well-formed expressions. These ill-formed expressions run the gamut of conceptually tight expressions that correspond to terms in a given ontology (e.g. relations labeled as "has part") to human-readable but machine-goes-bonkus-because-of-badly-formed-expressions (e.g. "Decisionmaking use cases built on science, technology, policy components"... machine goes 'huh'?). Some nodes in the c-soup are semantically annotated on an experimental basis. Individual concepts may ultimately (not done in this experiment) be represented as subject-predicate-object triples. Data-to-decisions traceability maps may be constructed by ordering the triples. These triples and ordered-triple-sequences may then further be exposed to algorithms for machine-assisted discovery.

These ideas were first developed as part of an 2018 ESIP Lab Project and documented in one of the project deliverables (human-readable with lots of diagrams!).

All the concept maps on this page are interactive and you can choose to drill down into connected concepts. This may get challenging for larger concept maps that are squeezed into fix-sized frames. Should you run into this issue with the interactive schema below, use this full-sized schema that will open in your browser window. Use your browser's back button to retrace your steps back "up" the concept maps.


What we need from you

We have sketched out a hypothetical "data to decisions" pathway below. Pathways that reflect real-world solutions are likely to involve the many disciplines that ESIP members hail from. Contribute your expertise by way of drafting concept maps along spots in the pathway that you feel need shoring up!

2020-03-20.Concept map icons for the data-to-decisions demo path wiki page v1.1.png
In the section below titled "Starting from decisions and tracing back to data and technology", we have worked out an example of what a pathway might look like for a Chesapeake watershed resilience challenge.

Why we're doing this

  • Context awareness and knowing where you fit in the big picture. So that professionals like yourself are able to see how technology you develop may be traced to a real-world applied problem, or vice versa.
  • Context-aware knowledge discovery. So that the community can start experimenting with using natural language processing methods to parse human-readable concept maps. Such maps are not as machine-readable as, say, an ontology crafted in OWL. However, such maps are an effective way to communicate with stakeholders, and can also be repurposed to inform machine-assisted discovery of data and information relevant to a real-world challenge. Queries with boolean operators, coupled with semantic technology, is great, but wouldn't it be even better if knowledge discovery algorithms reflect the inference and association patterns exhibited by human thinking that allows us to reason in the abstract?
  • Develop a resource that you can use to tell your own data-to-decisions story. You can pick and choose relevant concepts from the c-soup to communicate to other people (like your program manager!) about how data or technology you helped developed can ultimately be traced to science-informed, data-driven decisions. This you can do by assembling your own concept map embedded within a wiki, just like what we have done here on this wiki page. Or, use your concept map in any content management system that allows you to insert an HTML iframe into the body of your text. See the section below titled "Starting from decisions and tracing back to data and technology" for an example of a hypothetical "story" about a science-informed, data-driven resilience plan with elements that trace back to a NASA data product.


Real-world agriculture resilience scenario used to drive the concept map development

To anchor the "decision" end of the "data to decisions" pipeline, we shall focus on a use-case centered on climate adaptation challenges in the Chesapeake watershed. In particular, we focus on poultry agriculture in the Delmarva Peninsula (derived from the names of the States of Delaware, Maryland, and Virginia). The figure below shows USDA agriculture census data representing the value of poultry and eggs sold as a percentage of the total market value of agricultural products in 2017.

The hypothetical use-case involves an inter-county task force ("task force") charged with drafting a regional resilience plan to ameliorate the impacts of climate change on the economy and biodiversity of the Chesapeake watershed. First, the task force convened a multitude of stakeholders to derive a common understanding of the risks and vulnerabilities. Stakeholders included business owners, farmers, critical infrastructure operators, economic planners, scientists, environmental engineers, and government representatives. A shared understanding of the challenges facing the region was captured in a number of concept maps. These concept maps proved to be effective as visual communication tools to share the stakeholders' common understanding of the challenges confronting their communities.

Subsequently, the task force charged a team of transdisciplinary ESIP folks (yay!!!!) to develop a technology plan to support science-informed, data-driven decision-making. The ESIP folks were asked to first focus on meeting the needs of the poultry agriculture industry. However, the technology plan should also be scalable to meet the needs of other areas of concern like enhancing coastal flood abatement civil infrastructure, enhancing the long-term sustainability of Chesapeake oyster farms, etc.

Value of Poultry and Eggs Sold as Percent of Total Market Value of Agricultural Products Sold in 2017.png
(Image source: https://www.nass.usda.gov/Publications/AgCensus/2017/Online_Resources/Ag_Census_Web_Maps/, retrieved 2020-03-28 1615 hrs GMT.)

Starting from decisions and tracing back to data and technology

Click only on the orange-colored nodes as you navigate the c-soup. Strung sequentially together, these orange colored nodes constitute a pathway that brings the user through the science and technology landscape that crosses paths with decision-makers, informatics professionals, and domain scientists. Below is a high-level sequence of concepts (i.e. sequenced order of triples that capture the provenance of a decision) that was constructed using nodes in c-soup. Each node in the c-soup should ideally have its own Internationalized Resource Identifier (IRI) so that individual nodes from different concept maps can be chained to represent some process. This is NOT the case with the current prototype experimental implementation shown here.

The orange colored nodes below are clickable, but you would be left scrolling around madly within the restricted size window. Click here instead to render the full-scale concept map below in your browser window.
The high-level sequence of concepts above is an instantiated version of the diagram posted above (with apologies for the contra-directional flow): 2020-03-20.Concept map icons for the data-to-decisions demo path wiki page v1.1.png
The concept map below corresponds to the first orange node above. If you see nodes (equivalent terms: subject, object) or relations (equivalent term: predicate) with the expression ~sem.ann, those nodes or relations contain a semantic annotation. Mouse-over the node to see the annotation. Annotations are based on OBO Foundry ontologies like ENVO (Environment Ontology), AGRO (Agronomy Ontology), SDGIO (Sustainable Development Goals Interface Ontology), and RO (Relations Ontology). Please excuse the inconsistency in node annotation, because this is a phase of active experimentation. The experiment may be described as: "will semantic annotations of human-readable concepts (and perhaps predicates that connect concepts) enable these concept maps to be used for machine-assisted discovery of pertinent digital research objects?"

Click here to render the full-scale concept map below in your browser window.

To contribute and edit the c-soup

You can contribute to the c-soup by using either a desktop-based software called CmapTools, or the web-based version called CmapTools in the Cloud (henceforth referred to as "CmapToolsCloud". The desktop version, CmapTools, is recommended and available as freeware for the Windows, OSX, and Linux platforms.

To contribute and edit the c-soup, this is what you need to do at a conceptual level. You can find more detailed instructions right after the abstracted instructions:

  1. Create a free CmapToolsCloud account at https://cmapcloud.ihmc.us/index.
  2. Log into your CmapToolsCloud account, and add bwee --at-likethissign-@-- massiveconnections.com and/or william.l.teng --at-likethissign-@-- nasa.gov as a fellow Cmapper to your account.
  3. CmapToolsCloud will notify Brian Wee and/or Bill Teng to add you as a collaborator to the c-soup. Thereafter, you will be able to edit the c-soup.
  4. Download CmapTools from https://cmap.ihmc.us/cmaptools/ (optional, but recommended).
  5. Provide your CmapToolsCloud login account information to CmapTools so that you can edit the c-soup using CmapTools (optional, but recommended).


Here are the more detailed instructions:

  1. Create a free CmapToolsCloud account at https://cmapcloud.ihmc.us/index.
  2. Log into your CmapToolsCloud account.
  3. Find your Cmaps home page where you should see a tab labeled "Cmappers".
  4. Navigate to the page under the "Cmappers" tab, and click on "Find Cmappers".
  5. Look for either bwee --at-likethissign-@-- massiveconnections.com or william.l.teng --at-likethissign-@-- nasa.gov.
  6. Wait for a notification via email that you've been successfully added as a collaborator, which may take up to a day because it involves a manual action by the c-soup administrators. The email notification will be sent by cmap-no-reply@ihmc.us to the email address you use for CmapToolsCloud.
  7. (Following steps are optional, but recommended) Download CmapTools from https://cmap.ihmc.us/cmaptools/.
  8. Fire up CmapTools on your desktop, navigate your way to the "Cloud Account" setup (under "Edit" - "Preferences" in the Windows version), and specify your CmapToolsCloud login information.
  9. Once the c-soup administrators grant you collaboration access, you should now be able to contribute to the knowledge repository named the "AgClimate Concept Map Schema" using CmapTools or CmapToolsCloud.