[esip-semantictech] [EXTERNAL] Re: An ethnographic examination of some semantics
Mark Parsons
parsonsm.work at icloud.com
Fri May 28 18:22:33 EDT 2021
Yeah, what Thomas said. I don’t think Chris is being heretical either. Ruth and I are currently help apply heresy #1 on a project for NASA SMD. Heresy #2 is discussed a bit in the paper.
Few would argue that your search interface facets should exactly mirror your ontology/taxonomy. As I understand it, the point is for the underlying knowledge graph t(or hierarchy) to enable system function and interoperability via inference and other mechanisms. The user gets to call things what they want, how they want (within reason). Looking at just a few of the many GCMD keyword 'adoptions', there is quite a lot of variance in what is presented to the user.
I guess one question is what does “adoption” really mean in this case.
cheers,
-m.
> On May 28, 2021, at 2:02 PM, Huang, Thomas (US 398F) <thomas.huang at jpl.nasa.gov> wrote:
>
> I agree with Chris. The strength of semantic web is in inference. However, the deeper the inference the harder for our users to find the right result. IMHO, in order for semantic inference to be effective, it needs to be combined with search metrics and dynamic result ranking (i.e. search relevance).
>
> Thanks
>
> Thomas
>
> ----
> Thomas Huang
> Group Supervisor
> Data Product Generation Software
> Instrument Software and Science Data Systems
> Jet Propulsion Laboratory, California Institute of Technology
> (818) 354-2747 <tel:(818)%20354-2747>
>
>> On May 28, 2021, at 12:44 PM, Lynnes, Christopher S. (GSFC-5860) via esip-semanticweb <esip-semanticweb at lists.esipfed.org> wrote:
>>
>>
>> I can’t explain the overall SWEET non-uptake by data centers, but I can explain why I have usually ended up with something else (after trying SWEET) in Search interface use cases, such as in Usage-Based Discovery, If the community doesn’t mind a couple heretical statements…
>>
>> TL;DR: While SWEET may be a logical ontology from a knowledge organization standpoint, it does not fit well when you turn it around to use in a data search User Experience.
>>
>> Heresy #1: Currently, I build data system components by starting with the desired user experience and working downward to the application and the database+schema. The database/schema is whatever it needs to be to help the application satisfy the UX. (In 1991, I went in the opposite dierction, starting by modeling the data, building a UI on top of that. Results were, um, uneven.)
>>
>> It is more helpful for the UX to use terminology that “makes sense” to the end user, and in the context of the particular use case.
>>
>> Meresy #2: Multi-level hierarchies (SWEET, GCMD, or ANZSRC) are problematic in this sense: if you just slice through at one level to present to the user, some categories are too coarse and others are too fine in different places in order to work in a Search tool. To “make sense” to the user, we end up picking one term from the 3rd level, another from the 2nd level, another from the 4th...ugh.
>>
>> Alternatively, if we throw up our hands and make the user navigate the hiearchy on their own, the user has to guess at which branch to follow to find the thing they really want. If looking for “fire”, the user needs to infer that it is somewhere under phenReaction (Reaction Phenomena). Which an expert might recognize as the right place to find it, but not a regular person. (This is even more problematic with GCMD where very similar low-level terms are found in completely different branches.)
>>
>> I don’t mean to criticize SWEET, just to point out why it is hard to utilize in certain use cases. YMMV.
>> —
>> Christopher Lynnes NASA/GSFC mobile: 410-231-4573
>> “Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” A. de St. Exupery
>>
>> On 5/28/21, 14:28, "Mark Parsons" <parsonsm.work at icloud.com <mailto:parsonsm.work at icloud.com>> wrote:
>>
>> Thanks SJ and Chris. Definitely some refinement needed if we do submit somewhere. I’ve been amazed at folks's ability to dig up ancient gray lit. :-)
>>
>> BTW the human dimensions keywords are still there and even have been expanded (should probably note that). At the time of IPY most social scientists I talked to didn’t find them very helpful or in good alignment with what they were already using.
>>
>> Also I’d welcome any feedback on why SWEET wasn’t broadly adopted. My impression is that at first, most data centers were not really ready to think about ontologies and then when Rob died, they lost further momentum. That does appear to be changing as at least elements of SWEET are being added to ENVO and the OBO Foundry. Y’all probably know more.
>>
>> cheers,
>>
>> -m.
>>
>>
>> On May 28, 2021, at 12:20 PM, Siri Jodha Khalsa via esip-semanticweb <esip-semanticweb at lists.esipfed.org <mailto:esip-semanticweb at lists.esipfed.org>> wrote:
>>
>> Nice. Thanks, Chris, for digging out this detailed history. - sjs
>> On 5/28/21 7:17 PM, Lynnes, Christopher S. (GSFC-5860) wrote:
>> For the origin story of GCMD and its roots in the NASA Master Directory (ca. 1988) and IDN (ca. 1990), see: https://idn.ceos.org/newsletter/newsletter15.html <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=https*3A*2F*2Fidn.ceos.org*2Fnewsletter*2Fnewsletter15.html&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358937034*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=EHdrLl*2FSDrf0h8HNd7iZcQxrCXLUxISHr6yLJx1*2B1lM*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSU!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82_U3F67Y$>
>>
>> —
>> Christopher Lynnes NASA/GSFC mobile: 410-231-4573
>> “Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” A. de St. Exupery
>>
>> On 5/28/21, 12:13, "esip-semanticweb on behalf of Siri Jodha Khalsa via esip-semanticweb" <esip-semanticweb-bounces at lists.esipfed.org <mailto:esip-semanticweb-bounces at lists.esipfed.org> on behalf of esip-semanticweb at lists.esipfed.org <mailto:esip-semanticweb at lists.esipfed.org>> wrote:
>>
>> Hello Mark, Ruth and Øystein -
>> Thanks for a well-reasoned, somewhat provocative essay. It deserves to get widely read. An appropriate journal doesn't leap to mind, but I suspect others will have suggestions.
>> I was working in the EOSDIS Core System Science Office before launch of the first EOS satellite and from what I recall GCMD and EOSDIS began and evolved somewhat in parallel. The motivation for GCMD was to "facilitate interactions among science data cataloging systems" and it was the Committee on Earth Observing Satellites Data Working Group (CEOS-DWG) that sponsored the initial implementation of the International Directory Network (IDN) which became the GCMD. I may not have the exact details correct, but the statement "The GCMD was developed to support the launch of ... EOS." is probably incorrect.
>> Also of interest, and perhaps worth noting, is that initially the GCMD had "Socioeconomic" keywords contributed by CIESIN. These included, e.g.
>> Population Dynamics > Demographic Characteristics > ethnicity
>> Human Attitudes, Preferences, and Behavior > Social Behavior > resistance to change
>> Human and Environmental Health > Physiological Parameters > digestive physiology
>> Environmental Protection > Biota > plankton
>> which might have covered some of the needs in IPY, but I don't know what happened to these keywords. Also, note that originally, and for many years, there were only what is now called "science keywords" but at that time was called the "GCMD Parameter Valids".
>> In terms of "lessons learned", it might be worth discussing the reasons behind the observation that "SWEET has also not been broadly embraced by Earth science data centers as a standard ontology or vocabulary service."
>> If and when you do submit the paper to a journal, you're welcome to suggest me as a reviewer.
>> Best wishes,
>> SiriJodha
>>
>> On 5/27/21 4:07 PM, Mark Parsons via esip-semanticweb wrote:
>> Dear Semanticists,
>>
>> Some of you may be interested in this ethnographic tale of the development and evolution of the GCMD keywords.
>>
>> We would welcome any feedback on the essay, including possible journal suggestions for where it could be submitted.
>>
>> Parsons, Mark A., Duerr, Ruth, & Godøy, Øystein. (2021). The Evolution of GCMD Keywords — An instructive tale of data standards development and adoption. http://doi.org/10.5281/zenodo.4818237 <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=http*3A*2F*2Fdoi.org*2F10.5281*2Fzenodo.4818237&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358937034*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=wYAw3E8Wew8uhIIE0C8Vj*2F0mGCHuMqc6pow4c*2FGE4BQ*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSU!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82nTZZl5s$>
>>
>> Abstract
>> NASA established the Global Change Master Directory (GCMD) and supporting keywords in the early 1990s as part of implementing the GCMD through the Directory Interchange Format or DIF. The GCMD was developed to support the launch of the huge and enduring satellite-based Earth Observing System (EOS). The primary intent was to catalog EOS and related data, but the keywords have been implemented in many different systems and adopted in varying ways by many different organizations around the world. This essay provides an ethnographic examination of how the keywords have evolved and been managed and how they have been adopted over the last few decades. It illustrates how semantic approaches have evolved over time and provides insights on how standards and associated processes can be sustained and adaptable. Ongoing institutional commitment is essential, but so is transparency and technical flexibility. Understanding the different roles involved in standards creation, maintenance, and use of standards as well as the services that standards enable is also critical. It is apparent that semantic representations need to be mindful of different contexts and carefully define verbs as well nouns and categories. Understanding and representing relationships is central to interdisciplinary interoperability.
>>
>>
>> cheers,
>>
>> -m.
>>
>>
>> Mark A. Parsons
>> Research Scientist
>> University of Alabama in Huntsville
>> +1 303 941 9986
>> Mail: 1550 Linden Ave, Boulder, CO 80304, USA
>> https://orcid.org/0000-0002-7723-0950 <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=https*3A*2F*2Forcid.org*2F0000-0002-7723-0950&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358946990*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=q08p1kSm*2FNK29PZfithjA7ivWisFWIy2WWdBjvyXOec*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUl!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82Vl5K5Oo$>
>>
>>
>> _______________________________________________
>> esip-semanticweb mailing list
>> To start a new topic: esip-semanticweb at lists.esipfed.org <mailto:esip-semanticweb at lists.esipfed.org>
>> To unsubscribe and manage prefs: https://lists.esipfed.org/mailman/listinfo/esip-semanticweb <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=https*3A*2F*2Flists.esipfed.org*2Fmailman*2Flistinfo*2Fesip-semanticweb&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358946990*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=bCuYUQEuqJnCn06OSs2nZkc5*2F3IxwDwD*2F7U8P4t8ngw*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUl!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82JyS14wA$>
>> --
>> Siri-Jodha Singh KHALSA, Ph.D., SMIEEE
>> National Snow and Ice Data Center
>> University of Colorado
>> Boulder, CO 80309-0449
>> Traditional Territories of the Arapaho, Cheyenne and Ute Nations
>> Office: 1-303-492-1445 GV: 1-303-736-9976 EU: +420 608 720 281
>> http://orcid.org/0000-0001-9217-5550 <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=http*3A*2F*2Forcid.org*2F0000-0001-9217-5550&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358956946*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=vJ5s*2FigG66Hwe1hEuiFcOQlKDBvGX*2BVGGi*2FQ2PFqwTg*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSU!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu827ZuYKQc$>
>> --
>> Siri-Jodha Singh KHALSA, Ph.D., SMIEEE
>> National Snow and Ice Data Center
>> University of Colorado
>> Boulder, CO 80309-0449
>> Traditional Territories of the Arapaho, Cheyenne and Ute Nations
>> Office: 1-303-492-1445 GV: 1-303-736-9976 EU: +420 608 720 281
>> http://orcid.org/0000-0001-9217-5550 <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=http*3A*2F*2Forcid.org*2F0000-0001-9217-5550&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358956946*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=vJ5s*2FigG66Hwe1hEuiFcOQlKDBvGX*2BVGGi*2FQ2PFqwTg*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSU!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu827ZuYKQc$>
>>
>> _______________________________________________
>> esip-semanticweb mailing list
>> To start a new topic: esip-semanticweb at lists.esipfed.org <mailto:esip-semanticweb at lists.esipfed.org>
>> To unsubscribe and manage prefs: https://lists.esipfed.org/mailman/listinfo/esip-semanticweb <https://urldefense.us/v3/__https://gcc02.safelinks.protection.outlook.com/?url=https*3A*2F*2Flists.esipfed.org*2Fmailman*2Flistinfo*2Fesip-semanticweb&data=04*7C01*7Cchristopher.s.lynnes*40nasa.gov*7C8c67afa8c03f4d05926e08d922067294*7C7005d45845be48ae8140d43da96dd17b*7C0*7C0*7C637578233358966907*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C1000&sdata=SjXPgHuR3lCjI*2FrQdHl3sSwhsVEaTv*2BmmJvriTw3bY4*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUl!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82KyhHHFU$>
>>
>>
>>
>> _______________________________________________
>> esip-semanticweb mailing list
>> To start a new topic: esip-semanticweb at lists.esipfed.org <mailto:esip-semanticweb at lists.esipfed.org>
>> To unsubscribe and manage prefs: https://urldefense.us/v3/__https://lists.esipfed.org/mailman/listinfo/esip-semanticweb__;!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82IHQpGN0$ <https://urldefense.us/v3/__https://lists.esipfed.org/mailman/listinfo/esip-semanticweb__;!!PvBDto6Hs4WbVuu7!ZG_ls4KcStOMTL-Tl4THD3a_ohLG7_2mVFFRPqwtOrnVidXu8Xp5qlfIjnwLuu82IHQpGN0$>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.esipfed.org/pipermail/esip-semanticweb/attachments/20210528/e86ddcf6/attachment-0001.htm>
More information about the esip-semanticweb
mailing list