[Humanist] 28.842 an arguent & belief system

Humanist Discussion Group willard.mccarty at mccarty.org.uk
Sun Mar 22 08:26:09 CET 2015


                 Humanist Discussion Group, Vol. 28, No. 842.
            Department of Digital Humanities, King's College London
                       www.digitalhumanities.org/humanist
                Submit to: humanist at lists.digitalhumanities.org



        Date: Sat, 21 Mar 2015 09:07:08 +0000
        From: Dominic Oldman <doint at oldman.me.uk>
        Subject: Re:  28.819 an argument & belief system
        In-Reply-To: <20150315072307.9A021C29 at digitalhumanities.org>


Dear Martin,

I have put together some general notes together. I hope they are helpful.

CRMInf is a specialisation of the core CIDOC CRM ontology. The CRM provides a set of entities and relationships with precise real world scientific definitions that provide generalisations designed to harmonise structured data. It does not mandate fields and values and supports any structured information retaining its original meaning and perspectives. It is event based.

The Core CRM ontology provides the foundational classes and properties and is extremely robust to ensure that the ontology cannot be broken, to integrate different levels of knowledge, and support a wide range of potential specialisations derived from core classes. Extensions as opposed to specializations are increasingly rare.

CRMInf is a specialisation of the CRM derived from the abstraction of different argumentation theories. See http://dl.acm.org/citation.cfm?id=1921615. It is also worth saying up front that CRMInf is closely aligned with CRMSci which is a scientific observation extension of the CRM. 

The argumentation system currently works on the basis of three types of argument. 
1.    Observation,
2.    Inference and 
3.    Belief adoption.

The full reference and scope notes of all the labels can be found in the CRMInf draft athttp://www.ics.forth.gr/isl/index_main.php?l=e&c=713

These arguments operate on beliefs which have proposition sets. 

For example, in my domain an artefact has things written about it by the museum that physically owns it. These things are, in fact, part of an argument based on observation, since the museum owns and has access to an object in order to create documentation. However, anyone physically observing an object themselves can equally make an observation (an argument) about it. All beliefs can be attributed and other information is added that would allow a logical assessment of the argument.

Museum documentation may take the form of a database record and this record is a proposition. Within that proposition there may be things that people agree or disagree with, in which case they can adopt (belief adoption) parts of the proposition and challenge others (using any of the argument types). They may use the beliefs, along with beliefs collected from other sources in the community (and this is crucial), perhaps a citation or a different type of data resource, to construct an opposing argument. It is possible of course that the museum may, in time, adopt the arguments of researchers who have challenged it – in which case they would adopt the whole provenance of the arguments behind that belief. 

Another example, might be an interpretation. For example, an interpretation of a painting, perhaps from an owning institution, may form a belief or set of beliefs that can be challenged by someone else. These new beliefs could be used as premises and may include historical information that provides context for the argument, e.g. facts about the people and things depicted or more general historical facts from the period or place. This is information that can be adopted from many different sources to develop a conclusion that is also in the form of a belief – a new belief. Equally others may use one or more of these beliefs (adoption), as well as their own, and formulate their own arguments, all of which are connected and eventually lead back to the original premise or belief again forming a provenance of different assertions. 

Other examples recently used come from archaeology, for example using logic about the way different context (stratigraphic) can support an hypothesis about a find. 

No honest argument or belief is out of scope and many different types of inference logic can be used (philosophical, mathematical or other kinds of evidence), although some logic might be more weighty than others. This means that arguments are not lost and gaps do not develop over time that might lead to wrong conclusions. It also means that these arguments can be reasoned over using computers.

Anyway, taking the very simple initial example using the CRM ontology. A database record is a type of Information object (an E31 Document in CRM terms) – a document is a special type of information object and can be harmonised with other documents or indeed other information objects. For example, an information object is a special type of symbolic or propositional object which in turn are conceptual objects. I add this as an aside because CIDOC CRM is a poly-hierarchical system of meaning allowing different levels of knowledge to be integrated because, of course, people have different levels of knowledge which could not be integrated otherwise!   

In CRMInf this particular type of information object is called a proposition set (I4 Proposition Set). It is therefore both a document and a proposition set. A proposition is also a type of information object and, “comprises identifiable symbols and any aggregation of symbols”.   In other words it can be text, images, musical scores, emblems, signs, and so on. This is important because it means that different types of information, heterogeneous data, can be connected without conversion to a single knowledge representation system. However, the CRM itself is about information/knowledge harmonisation and CRMInf becomes extremely powerful when the information that it operates on is encoded in the same knowledge representation system - but CRMInf is not limited by doing this first.

This proposition forms the basis of a belief. A belief consists of a proposition and a belief value. The value is whatever is appropriate. In this case let’s say, True, False, Unknown. The museums belief was created as a result of an observation.
If someone was working with an image of an object then the proposition that they start with is different and based on the production of that image which was part of a previous observation.  This is an important distinction and may affect the way that an argument is perceived and weighted – particularly internet research in which people sometimes talk about an object as if they have physically examined it.     

An inference making argument is different in that it can use beliefs that may have resulted from observations or adoptions. Inference making usually uses as a premise an existing belief and can, with other premises conclude a new belief. Belief adoption takes an existing belief and makes it yours. The adoption concludes in your own belief.   

In CIDOC CRM terms…. 

S4 Observation -> J2 concluded that -> I2 Belief -> J4 that -> I4 Proposition set AND J5 holds to be -> I6 Belief Value. (will stop at belief after this)

e.g. An Observation activity/event  (argument) concluded a belief that is made up of a proposition and a value. 

I5 Inference Making ->   J1 used as a premise -> I2 Belief

e.g. An Inference making activity/event (argument) used a belief as a premise.

I7 Belief Adoption -> J6 Adopted -> I2 Belief

e.g. A belief adoption activity/event resulted in the adoption of an existing belief 

I7 Belief Adoption -> J2 concluded that -> I2 Belief

e.g., The belief adoption activity/event therefore concluded in a belief 

I5 Inference making -> J3 applies -> I3 Inference logic

e.g. The inference making activity/event applied a logic 

I5 Inference making -> J2 concluded that -> I2 Belief

e.g. The inference making activity/event concluded in a new belief 

In other words. I can take a belief and use it as a premise. I can also use my own beliefs derived from my own observations and adopt other beliefs from different sources of information and make these my beliefs. I can then through the activity of inference making apply a logic from which I conclude, most likely, a set of beliefs that negate some of the original premises (beliefs) and propose new beliefs.

Anyway, we are currently in the middle of incorporating this into a software application. I will put the designs up on the Web soon.

Cheers,

Dominic





More information about the Humanist mailing list