Issue 510: belief adoption

Starting Date: 
Working Group: 

Posted by Thomas Bottini on 6/7/2020

Dear all,

We try to use CRMinf to model a scientific controversy about the attribution of a museum item (the Marie-Antoinette’s travel kit).

We would like to express the fact that a researcher adopts a belief (I7 Belief Adoption) after having studied the item at the museum (S4 Observation).

Why can’t the range of a J7 (is based on evidence from) be a S4 Observation (meaning a E7 Activity)? 

In our case, we don’t have any evidence of E73 (Information Object) type, the observation activity carried out by the researcher IS the evidence.

Thank you very much, in advance,



Current Proposal: 

Posted by George on 6/7/2020

Dear Thomas,

As I would read it, S4 Observation is a subclass of I1 Argumentation, therefore inheriting all of its properties. This being the case, an observation can lead an actor involved in it to come to conclude in a belief (J2). Therefore if the situation is that the scientist goes and analyzes the object (instance of S4) looking at certain properties, and then comes to some sort of belief, then this belief can be documented using J2 concluded that I2 Belief and then continue from there.

Belief adoption, to my understanding, should be used when the belief that one is taking up is not founded in one's own observational acts, but is rather simply taken over from some external authority. Therefore, you would not need two events, the observing, and the belief adopting. Rather you would need one event, the observation, which directly leads to a belief state.

Without any further context, that is how I imagine it should be modelled. CRMinfers, do I have it right?


Posted by Martin on 6/07/2020

Absolutely! "Belief Adaption" means "adopt another one's belief.

Whatever is found on a physical thing is an observation by human senses or other instruments receiving signals, including from chemical reactions, x-ray reflection and transmission, tactile etc.

There may be non-trivial Inferences subsequent to primary observation. For instance, abrasions at amphora handles regarded to stem from ropes that tied cargo in a ship.

Some instruments contain firmware that cannot be separated from the primary signal. We regard then the result as the primary observation, having in mind how the instrument works.


Posted by Olivier Marlet on 8/7/2020

For the logicist publication of the Rigny archaeological excavations, we used the CRMinf to model the principle of logicist argumentation according to Jean-Claude Gardin, which is rather convenient since the CRMinf is directly inspired by this theory.
In our case, we have distinguished 3 processes: 1/ argumentation based on observation or comparison data; 2/ external reference data (what is known and acquired elsewhere, taken from a bibliographical source for example); 3/ arguments built from previous conclusions.


1/ For a proposition based on observation data or comparison data, mapping could be:

S15_Observable_Entity → O11_was_described_by → S6_Data_evaluation (IsA I5_Inference_Making IsA I1_Argumentation) → J2_conclued_that → I2_Belief → J4_that → I4_Proposition_Set

I5_Inference_Making → J3_applies → I3_Inference_Logic


2/ For a proposition based on reference data, mapping could be:

E31_Document (IsA E73_Information_Object) → J7_is_evidence_for → I7_Belief_Adoption (IsA I1_Argumentation) → J6_adopted → I2_Belief → J4_that → I4_Proposition_Set


3/ For intermediate or final propositions, mapping could be:

I4_Proposition_Set → J4_is_subject_of → I2_Belief → J1_was_premise_for → S8_Categorical_hypothesis_building (IsA I5_Inference_Making IsA I1_Argumentation) → J2_conclued_that → I2_Belief → J4_that → I4_Proposition_Set    

I invite you to read our online article :

and to consult the resulting online publication in TEI format:
Here is the schema that helps me to better understand the organization of the CRMinf.
Hope it will be useful.

Posted by Thomas Bottini on 8/07/2020

It appears that I misread the CRMinf documentation, and thought that every I2 Belief should be associated to a I7 Belief Adoption. I was not able to deduce from the scope notes of I7 that it " is the acceptance of somebody else's conclusion about some state of affairs". Stephen's wording is extremely clear.

And thank you George for pointing out that S4 is a subclass of I1.

This leads to the very simple pattern: S4 ---[J2]---> I2

Olivier, thank you very much for the wonderful conceptual and graphical resources you have posted. They will be very useful for our further work.

Posted by Athina on 8/7/2020

I am wondering about the example of I7 Belief Adoption "My adoption of the belief that Dragendorff type 29 bowls are from the 1st Century AD". Maybe, it should be rephrased in order to express more precisely the trust in the source (which is someone else's) and in this sentence and it is actually implied.
just a thought,

Posted by Martijn Van Leusen on 8/7/2020

Thank you Olivier for your clear explanation of the three 'routes'! Maybe the following example of belief adoption could be useful: My source document (Vittucci 1968: 21) has interpreted a particular set of field observations as evidence for the presence of a roman farmstead; trusting in her ability to recognise this type of site, I adopt her belief.

Posted by Martin on 8/7/2020

The Scope Note of I7 will be corrected.

Posted by Martin on 8/7/2020

On 7/8/2020 1:42 PM, athinak wrote:

Dear all,

I am wondering about the example of I7 Belief Adoption "My adoption of the belief that Dragendorff type 29 bowls are from the 1st Century AD". Maybe, it should be rephrased in order to express more precisely the trust in the source (which is someone else's) and in this sentence and it is actually implied.
just a thought,

Yes, examples should also be updated!

Posted by Olivier Marlet on 8/7/2020

The notions of trust and adoption of beliefs are very interesting because they are directly related to the Re-use of FAIR principles. It is certainly a notion on which the ARIADNEplus working group will work in the sub-task 4.4.12 "CIDOC-CRM mapping for Excavation archives" dealing with the link between data and publications.

Posted by Franco on 8/7/2020

I wrote on this topic a paper with Sorin Hermon, some time ago (2017)

"Expressing Reliability with CIDOC CRM", IJDL, 18(4),

It’s available from the IJDL, from the ACM DL, and from me as a self-stored paper.

In short, it shows how to deal with trust in the Re-use framework of FAIR without using CRMinf. The necessary concepts are just subclasses and subproperties of existing ones. They receive special names just for the sake of clarity, but they could just be typized e.g. (Z denotes the new classes):

Z1 Reliability Assessment = E16 Measurement + has type “reliability assessment"
Z2 Reliability = E54 Dimension + has type “reliability"

Sooner or later, I should re-examine the issue in light of the recent discussions. 

In my opinion, “trust" should be machine-actionable otherwise Re-use becomes a purely human activity, but an undoable one as we don’t have enough time to read everything and take the necessary decisions, discarding fake news, as Gardin stated 21 years ago (*). Trust is a chain and at some point one of the referees needs to be “a honourable man” (**), whose assessment is automatically applied to the data together, and if too low it automatically discredits the data and avoids re-use or warns against it.

There are other divertissements of mine on the topic I’ll gladly share with those interested.


(*) Gardin, J.-C. “Calcul et narrativité dans les publications archéologiques”, Archeologia e Calcolatori, 10, 1999, 63-78. Open access.
(**) as everybody knows, this quote actually referred to an untrustworthy person