Re: [Arches] uncertain information in Arches
things through and discuss. In our database we have a field
for certain the data entry person is about the location of the site, ie.
about the polygon they might have drawn on a map. This field only allowed 5
choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5
(I have no idea whatsoever where the site is). We had a very detailed
manual with examples of all these cases, what to use when, ...
Final result of our experiment: every group had entered the location with a
different level of certainty. So, based on the exact same information they
had all drawn totally different conclusions. And this was about something
as simple as the location of the site.
So, I'm very curious about how you manage to prevent stuff like this from
happening.
The other thing I wonder about: how does certainty affect searching? Should
a search for 'churches' only return sites that have a certain certainty
attached to the interpretation? Are you working with sliding scale of
certainty (ie. we are 75% percent certain about this statement) or a binary
one (we're certain or uncertain)?
Cheers,
Koen
Van: arches...@googlegroups.com (javascript:) [arches...@googlegroups.com
(javascript:)] namens dwut...@fargeo.com (javascript:) [dwut...@fargeo.com
(javascript:)]
Verzonden: donderdag 27 maart 2014 22:36
Aan: thomas@gmail.com (javascript:)
CC: arches...@googlegroups.com (javascript:)
Onderwerp: Re: [Arches] uncertain information in Arches
Thomas,
Good question! You are quite correct that we haven’t tried to include
uncertainty in Arches.
One reason is pretty basic: certainty is quite subjective from person to
person. For example: most people agree that the earth is spherical. But a
“flat-earther” may be very certain that the earth is not a sphere, but is
instead a plane. His certainty does not make him correct, it merely states
the degree to which he believes in his interpretation. Clearly, you can be
very certain and very wrong at the same time. I guess my point is that in
many cases “certainty” says more about the person making the assertion than
it does about the thing being described.
OK, all philosophy aside, one could easily extend any Arches graph to
include a “certainty node”. Such a node could point to a thesaurus (as
many of the nodes in Arches already do), allowing a user to select from a
list of “uncertainty levels”. Really, any Arches graph could include a
“certainty node” under any entity that you might want to qualify (for
example, one certainty node for period and another certainty node for
heritage type).
Really, the hard part is not in getting Arches to allow you to add an
“uncertainty level” to your cultural heritage data. Rather, the difficult
thing is to decide how you’ll get different people to agree on what
constitutes certain vs. uncertain interpretations of heritage.
Sorry that I can’t be any more helpful… However, I’m very interested to
hear how you will model uncertainty and how you will get people to
implement it consistently. Please keep me posted!
Cheers,
Dennis
On Mar 27, 2014, at 2:41, thomas@gmail.com
(javascript:)mailto:thomas@gmail.com (javascript:) wrote:
I have a question about conceptual modeling in CIDOC CRM, maybe there is
someone one the list who is able to provide some guidance.
As posted before, we are trying to integrate research data of neolithic
sites into Arches. Now, naturally a significant part of this data has a
level of certainty to which the information is correct. e.g. a site can
consist of some features for certain (in this case modeled in the
Archaeological Heritage (Site).E27 - Component.E18 relationship) but if
others exist is uncertain. We believe this valuable information should not
get lost (quite often theory construction is based on such information).
For example it could be uncertain if an archaeological feature is to be
named pit or ditch - or if it exists at all. Another example could be
the questionable relationship of a findspot to a certain archaeological
period. To make it even more difficult, different authors could have
different thoughts on that.
As far as we can see, the expression of such uncertainty is not covered
by Arches yet. Is there a concept for the integration of such data in the
future? We are currently thinking into potential solutions but are
struggeling to find adequate expressions for uncertain information in
CIDOC.
thanks, Thomas
--
-- To post, send email to arches...@googlegroups.com
(javascript:)mailto:arches...@googlegroups.com (javascript:). To
unsubscribe, send email to archesprojec...@googlegroups.com
(javascript:)mailto:archesproject
Re: [Arches] uncertain information in Arches
,i3mainz COMPONENT_CERTAINTY_3,3,sortorder,i3mainz Can anybody tell us what's wrong with our conceptid? Does the authority_files.py search for the ID in any additional place, where we should reference it? Cheers, Tobias On Monday, March 31, 2014 11:22:27 PM UTC+2, Koen Van Daele wrote: Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven’t tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a “flat-earther” may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases “certainty” says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a “certainty node”. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of “uncertainty levels”. Really, any Arches graph could include a “certainty node” under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an “uncertainty level” to your cultural heritage data. Rather, the difficult thing is to decide how you’ll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can’t be any more helpful… However, I’m very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:thomas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has a level of certainty to which the information is correct. e.g. a site can consist of some features for certain (in this case modeled in the Archaeological Heritage (Site).E27 - Component.E18 relationship) but if others exist is uncertain
Re: [Arches] uncertain information in Arches
,i3mainz Can anybody tell us what's wrong with our conceptid? Does the authority_files.py search for the ID in any additional place, where we should reference it? Cheers, Tobias On Monday, March 31, 2014 11:22:27 PM UTC+2, Koen Van Daele wrote: Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases certainty says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a certainty node. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of uncertainty levels. Really, any Arches graph could include a certainty node under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an uncertainty level to your cultural heritage data. Rather, the difficult thing is to decide how you'll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can't be any more helpful... However, I'm very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:thomas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has a level of certainty to which the information is correct. e.g. a site can consist of some features for certain (in this case modeled in the Archaeological Heritage (Site).E27 - Component.E18 relationship) but if others exist is uncertain. We believe this valuable information should not get lost (quite often theory construction is based on such information). For example it could be uncertain if an archaeological feature is to be named pit or ditch - or if it exists at all. Another example could be the questionable relationship of a findspot to a certain archaeological period. To make it even more difficult, different authors could have different thoughts on that. As far as we can see, the expression
Re: [Arches] uncertain information in Arches
COMPONENT_CERTAINTY_1,1,sortorder,i3mainz COMPONENT_CERTAINTY_2,2,sortorder,i3mainz COMPONENT_CERTAINTY_3,3,sortorder,i3mainz Can anybody tell us what's wrong with our conceptid? Does the authority_files.py search for the ID in any additional place, where we should reference it? Cheers, Tobias On Monday, March 31, 2014 11:22:27 PM UTC+2, Koen Van Daele wrote: Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases certainty says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a certainty node. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of uncertainty levels. Really, any Arches graph could include a certainty node under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an uncertainty level to your cultural heritage data. Rather, the difficult thing is to decide how you'll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can't be any more helpful... However, I'm very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:th omas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has a level of certainty to which the information is correct. e.g. a site can consist of some features for certain (in this case modeled in the Archaeological Heritage (Site).E27 - Component.E18 relationship) but if others exist is uncertain. We believe this valuable information should not get lost (quite often theory construction is based on such information). For example it could be uncertain if an archaeological feature is to be named pit or ditch - or if it exists at all. Another example could be the questionable relationship of a findspot to a certain archaeological period. To make it even
Re: [Arches] uncertain information in Arches
']) KeyError: 'CONCEPTID' Our provisional Authority Files look like this: - COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv conceptid,PrefLabel,AltLabels,ParentConceptid,ConceptType,Provider COMPONENT_CERTAINTY_1,certain,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz COMPONENT_CERTAINTY_2,uncertain,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz COMPONENT_CERTAINTY_3,unknown,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz - COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.values.csv (do we need this one?) conceptid,Value,ValueType,Provider COMPONENT_CERTAINTY_1,1,sortorder,i3mainz COMPONENT_CERTAINTY_2,2,sortorder,i3mainz COMPONENT_CERTAINTY_3,3,sortorder,i3mainz Can anybody tell us what's wrong with our conceptid? Does the authority_files.py search for the ID in any additional place, where we should reference it? Cheers, Tobias On Monday, March 31, 2014 11:22:27 PM UTC+2, Koen Van Daele wrote: Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases certainty says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a certainty node. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of uncertainty levels. Really, any Arches graph could include a certainty node under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an uncertainty level to your cultural heritage data. Rather, the difficult thing is to decide how you'll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can't be any more helpful... However, I'm very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:th omas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has
Re: [Arches] uncertain information in Arches
/arches-web/archesproject/packages/cdscert/install/authority_files.py, line 63, in load_authority_file concepts.insert_concept(settings.DATA_CONCEPT_SCHEME, adoc_dict['PREFLABEL'], '', 'en-us', adoc_dict['CONCEPTID']) KeyError: 'CONCEPTID' Our provisional Authority Files look like this: - COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv conceptid,PrefLabel,AltLabels,ParentConceptid,ConceptType,Provider COMPONENT_CERTAINTY_1,certain,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz COMPONENT_CERTAINTY_2,uncertain,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz COMPONENT_CERTAINTY_3,unknown,,COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.csv,Index,i3mainz - COMPONENT CERTAINTY TYPE AUTHORITY DOCUMENT.values.csv (do we need this one?) conceptid,Value,ValueType,Provider COMPONENT_CERTAINTY_1,1,sortorder,i3mainz COMPONENT_CERTAINTY_2,2,sortorder,i3mainz COMPONENT_CERTAINTY_3,3,sortorder,i3mainz Can anybody tell us what's wrong with our conceptid? Does the authority_files.py search for the ID in any additional place, where we should reference it? Cheers, Tobias On Monday, March 31, 2014 11:22:27 PM UTC+2, Koen Van Daele wrote: Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases certainty says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a certainty node. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of uncertainty levels. Really, any Arches graph could include a certainty node under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an uncertainty level to your cultural heritage data. Rather, the difficult thing is to decide how you'll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can't be any more helpful... However, I'm very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:th omas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe
Re: [Arches] uncertain information in Arches
certain or uncertain)? Cheers, Koen Van: arches...@googlegroups.com [arches...@googlegroups.com] namens dwut...@fargeo.com [dwut...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas@gmail.com CC: arches...@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven’t tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a “flat-earther” may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases “certainty” says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a “certainty node”. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of “uncertainty levels”. Really, any Arches graph could include a “certainty node” under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an “uncertainty level” to your cultural heritage data. Rather, the difficult thing is to decide how you’ll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can’t be any more helpful… However, I’m very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas@gmail.commailto:thomas@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has a level of certainty to which the information is correct. e.g. a site can consist of some features for certain (in this case modeled in the Archaeological Heritage (Site).E27 - Component.E18 relationship) but if others exist is uncertain. We believe this valuable information should not get lost (quite often theory construction is based on such information). For example it could be uncertain if an archaeological feature is to be named pit or ditch - or if it exists at all. Another example could be the questionable relationship of a findspot to a certain archaeological period. To make it even more difficult, different authors could have different thoughts on that. As far as we can see, the expression of such uncertainty is not covered by Arches yet. Is there a concept for the integration of such data in the future? We are currently thinking into potential solutions but are struggeling to find adequate expressions for uncertain information in CIDOC. thanks, Thomas -- -- To post, send email to arches...@googlegroups.commailto:arches...@googlegroups.com. To unsubscribe, send email to archesprojec...@googlegroups.commailto:archesproject+unsubscr...@googlegroups.com. For more information, visit https://groups.google.com/d/forum/archesproject?hl=en --- You received this message because you are subscribed to the Google Groups Arches Project group. To unsubscribe from this group and stop receiving emails from it, send an email to archesprojec...@googlegroups.commailto:archesproject+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout. -- -- To post, send email to arches...@googlegroups.com
Re: [Arches] uncertain information in Arches
Hi Phil, thanks for the feedback! We totally agree that it doesn't make sense to model whether a site has finds or not. Our situation is a little different though, which I didn't express very clearly. In our case we are often rather unsure if a find or feature belongs to the phase that the site has been dated to, based on other finds/features. Two hypothetical examples: - Let's say we are sure that the ceramics found at site A belong to our interpretation of the site, but we are unsure if the burials that were found at the site do, as well. - Or we might be sure that one feature of the site is a post hole, but we are not sure that the other feature can be interpreted as a ditch or as a pit. Then we might try to model our uncertainty using the P3 (has note) - E62 (String) route or the P2 (has type) - E55 (Type) route as proposed by Geoff Carver in the Antiquist list: https://groups.google.com/forum/?hl=enfromgroups=#!topic/antiquist/w0R9VXDk5U4 Thanks again for any further thoughts! Best, Tobias On Monday, March 31, 2014 12:26:11 PM UTC+2, Phil Carlisle wrote: Hi Tobias, Thomas et al, It is perfectly valid to assign a level of certainty to an interpretation of a site/find. Thus a type assignment of PIT with a note of 'Uncertain' or a phase type assignment of 'Medieval' with a '?' is fine and indeed we (English Heritage) use something similar. However I noticed on the Antiquist list that Tobias has used the example of wanting to model the uncertainty of whether a site may have finds or not. Is that right or have I misunderstood? As far as I'm aware the CRM cannot model this situation and in fact would argue against modelling it at all. If you are certain that a site is an ARCHAEOLOGICAL SITE you can record it as such but if it is unexcavated and you are just conjecturing what it may contain then that should just be a note in the description. So I think that if you are modelling the uncertainty of an interpretation then you can use the P3-E62 route but you can't really model the uncertainty of the existence of a thing using the CRM. I hope this helps, and others may have different opinions! Phil *Phil Carlisle* Data Standards Supervisor Data Standards Unit, Designations Department English Heritage The Engine House Fire Fly Avenue Swindon SN2 2EH Tel: +44 (0)1793 414824 http://thesaurus.english-heritage.org.uk/ http://www.heritagedata.org/blog/ -- *From:* arches...@googlegroups.com javascript: [mailto: arches...@googlegroups.com javascript:] *On Behalf Of *Tobias Kohr *Sent:* 28 March 2014 14:33 *To:* arches...@googlegroups.com javascript: *Cc:* thomas@gmail.com javascript: *Subject:* Re: [Arches] uncertain information in Arches Hi Dennis, thanks for your quick response! Working together with Thomas, in fact one option that came to our mind was to attach certainty nodes to the entities where we need them (Component.E18 and Phase Type Assignment.E17 in our case) to technically implement uncertainty in Arches. We are struggling, however, to find an entity type in CIDOC CRM that seems adequate for modelling certainties. The only type that seems a possibility to us here is E59 Primitive Value / E62 String. So, conceptually we ask ourselves if there is a more adequate CIDOC type (which is designed for modelling certainties). And philosophically we are not sure if it doesn't contradict the idea of a E18 Physical Thing (Component.E18) to possess an attribute that expresses (un-)certainty of existence. We appreciate any comments or hints on these thoughts! Best, Tobias On Thursday, March 27, 2014 10:36:05 PM UTC+1, Dennis Wuthrich wrote: Thomas, Good question! You are quite correct that we haven’t tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a “flat-earther” may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases “certainty” says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a “certainty node”. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of “uncertainty levels”. Really, any Arches graph could include a “certainty node” under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you
RE: [Arches] uncertain information in Arches
Hi Tobias, Yes, Geoff's option is equally valid to the P3.E62 and obviously allows you to type the uncertainty. As I think Dennis mentioned the addition of certainty types to any of the nodes (following the P2.E55 route) is perfectly valid (with the addition of a -P71 lists (is listed in) E32.Uncertainty Authority Document). Phil Phil Carlisle Data Standards Supervisor Data Standards Unit, Designations Department English Heritage The Engine House Fire Fly Avenue Swindon SN2 2EH Tel: +44 (0)1793 414824 http://thesaurus.english-heritage.org.uk/ http://www.heritagedata.org/blog/ From: archesproject@googlegroups.com [mailto:archesproject@googlegroups.com] On Behalf Of Tobias Kohr Sent: 31 March 2014 14:52 To: archesproject@googlegroups.com Cc: 'Tobias Kohr'; thomas.enge...@gmail.com Subject: Re: [Arches] uncertain information in Arches Hi Phil, thanks for the feedback! We totally agree that it doesn't make sense to model whether a site has finds or not. Our situation is a little different though, which I didn't express very clearly. In our case we are often rather unsure if a find or feature belongs to the phase that the site has been dated to, based on other finds/features. Two hypothetical examples: - Let's say we are sure that the ceramics found at site A belong to our interpretation of the site, but we are unsure if the burials that were found at the site do, as well. - Or we might be sure that one feature of the site is a post hole, but we are not sure that the other feature can be interpreted as a ditch or as a pit. Then we might try to model our uncertainty using the P3 (has note) - E62 (String) route or the P2 (has type) - E55 (Type) route as proposed by Geoff Carver in the Antiquist list: https://groups.google.com/forum/?hl=enfromgroups=#!topic/antiquist/w0R9VXDk5U4 Thanks again for any further thoughts! Best, Tobias On Monday, March 31, 2014 12:26:11 PM UTC+2, Phil Carlisle wrote: Hi Tobias, Thomas et al, It is perfectly valid to assign a level of certainty to an interpretation of a site/find. Thus a type assignment of PIT with a note of 'Uncertain' or a phase type assignment of 'Medieval' with a '?' is fine and indeed we (English Heritage) use something similar. However I noticed on the Antiquist list that Tobias has used the example of wanting to model the uncertainty of whether a site may have finds or not. Is that right or have I misunderstood? As far as I'm aware the CRM cannot model this situation and in fact would argue against modelling it at all. If you are certain that a site is an ARCHAEOLOGICAL SITE you can record it as such but if it is unexcavated and you are just conjecturing what it may contain then that should just be a note in the description. So I think that if you are modelling the uncertainty of an interpretation then you can use the P3-E62 route but you can't really model the uncertainty of the existence of a thing using the CRM. I hope this helps, and others may have different opinions! Phil Phil Carlisle Data Standards Supervisor Data Standards Unit, Designations Department English Heritage The Engine House Fire Fly Avenue Swindon SN2 2EH Tel: +44 (0)1793 414824 http://thesaurus.english-heritage.org.uk/ http://www.heritagedata.org/blog/ From: arches...@googlegroups.comjavascript: [mailto:arches...@googlegroups.comjavascript:] On Behalf Of Tobias Kohr Sent: 28 March 2014 14:33 To: arches...@googlegroups.comjavascript: Cc: thomas@gmail.comjavascript: Subject: Re: [Arches] uncertain information in Arches Hi Dennis, thanks for your quick response! Working together with Thomas, in fact one option that came to our mind was to attach certainty nodes to the entities where we need them (Component.E18 and Phase Type Assignment.E17 in our case) to technically implement uncertainty in Arches. We are struggling, however, to find an entity type in CIDOC CRM that seems adequate for modelling certainties. The only type that seems a possibility to us here is E59 Primitive Value / E62 String. So, conceptually we ask ourselves if there is a more adequate CIDOC type (which is designed for modelling certainties). And philosophically we are not sure if it doesn't contradict the idea of a E18 Physical Thing (Component.E18) to possess an attribute that expresses (un-)certainty of existence. We appreciate any comments or hints on these thoughts! Best, Tobias On Thursday, March 27, 2014 10:36:05 PM UTC+1, Dennis Wuthrich wrote: Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his
RE: [Arches] uncertain information in Arches
Hi all, I just wanted to get back at what Dennis said at the beginning of this thread. Im quite curious how you will get people to agree on (un)certainty. If feels like a very natural idea to talk and think about, but I haven't really seen it function properly in practice. We once did an experiment where we had 10 people who were used to entering data in our archaeological inventory system enter the same site. We paired the archaeologists: one more more experienced data entry person (a few years experience) and one newbie (a few months), so they would be forced to really think things through and discuss. In our database we have a field for certain the data entry person is about the location of the site, ie. about the polygon they might have drawn on a map. This field only allowed 5 choices, ranging from 1 (I'm sure it's exactly where it needs to be) to 5 (I have no idea whatsoever where the site is). We had a very detailed manual with examples of all these cases, what to use when, ... Final result of our experiment: every group had entered the location with a different level of certainty. So, based on the exact same information they had all drawn totally different conclusions. And this was about something as simple as the location of the site. So, I'm very curious about how you manage to prevent stuff like this from happening. The other thing I wonder about: how does certainty affect searching? Should a search for 'churches' only return sites that have a certain certainty attached to the interpretation? Are you working with sliding scale of certainty (ie. we are 75% percent certain about this statement) or a binary one (we're certain or uncertain)? Cheers, Koen Van: archesproject@googlegroups.com [archesproject@googlegroups.com] namens dwuthr...@fargeo.com [dwuthr...@fargeo.com] Verzonden: donderdag 27 maart 2014 22:36 Aan: thomas.enge...@gmail.com CC: archesproject@googlegroups.com Onderwerp: Re: [Arches] uncertain information in Arches Thomas, Good question! You are quite correct that we haven't tried to include uncertainty in Arches. One reason is pretty basic: certainty is quite subjective from person to person. For example: most people agree that the earth is spherical. But a flat-earther may be very certain that the earth is not a sphere, but is instead a plane. His certainty does not make him correct, it merely states the degree to which he believes in his interpretation. Clearly, you can be very certain and very wrong at the same time. I guess my point is that in many cases certainty says more about the person making the assertion than it does about the thing being described. OK, all philosophy aside, one could easily extend any Arches graph to include a certainty node. Such a node could point to a thesaurus (as many of the nodes in Arches already do), allowing a user to select from a list of uncertainty levels. Really, any Arches graph could include a certainty node under any entity that you might want to qualify (for example, one certainty node for period and another certainty node for heritage type). Really, the hard part is not in getting Arches to allow you to add an uncertainty level to your cultural heritage data. Rather, the difficult thing is to decide how you'll get different people to agree on what constitutes certain vs. uncertain interpretations of heritage. Sorry that I can't be any more helpful... However, I'm very interested to hear how you will model uncertainty and how you will get people to implement it consistently. Please keep me posted! Cheers, Dennis On Mar 27, 2014, at 2:41, thomas.enge...@gmail.commailto:thomas.enge...@gmail.com wrote: I have a question about conceptual modeling in CIDOC CRM, maybe there is someone one the list who is able to provide some guidance. As posted before, we are trying to integrate research data of neolithic sites into Arches. Now, naturally a significant part of this data has a level of certainty to which the information is correct. e.g. a site can consist of some features for certain (in this case modeled in the Archaeological Heritage (Site).E27 - Component.E18 relationship) but if others exist is uncertain. We believe this valuable information should not get lost (quite often theory construction is based on such information). For example it could be uncertain if an archaeological feature is to be named pit or ditch - or if it exists at all. Another example could be the questionable relationship of a findspot to a certain archaeological period. To make it even more difficult, different authors could have different thoughts on that. As far as we can see, the expression of such uncertainty is not covered by Arches yet. Is there a concept for the integration of such data in the future? We are currently thinking into potential solutions but are struggeling to find adequate expressions for uncertain information in CIDOC
