RE: Scientific Models and Semantics

Hi Soren



In the context of a recent RuleML paper<https://www.semanticscholar.org/paper/Towards-Supporting-Multiple-Semantics-of-Named-N3-Arndt-Woensel/1557bfec3746bb0c8b1209179f0a3af3884e85f1> on graph semantics, we computerized multiple contradicting medical guidelines (i.e., on the use of aspirin and its relation to Reye's syndrome) and presented different ways of stating their semantics. Perhaps this could be an interesting use case for ORKG - i.e., the machine interpretation of evidences allows a researcher to more easily identify "support" and "attack" relations between scientific publications.



That said, in the example above, any medical expert worth their degree would already know about the alleged connection between aspirin and Reye's syndrome (which is why currently, at least in Canada, aspirin is only recommended for heart failure risk). But I think the problem could be applied to other, lesser known issues and also be generalized towards other fields.





William



-----Original Message-----
From: Sören Auer <auer@l3s.de>
Sent: February-10-20 9:29 AM
To: semantic-web@w3.org
Subject: Re: Scientific Models and Semantics



Dear Adam, all,



I guess its not 100% related, but we are currently working on representing scientific contributions in a knowledge graph. The current beta version of the Open Research Knowledge Graph you can find there:

https://orkg.org



We aim to create structured comparisons of research approaches, such as e.g. this one for semi-supervised author name disambiguation:



https://www.orkg.org/orkg/comparison/R6187



Regarding your example, we envision, that different scientific models will then consist of different entities, i.e. there are different resources referring to the different conceptual models of an electron.



We are currently looking to create more examples from non-CS domains - please let us know if you have any ideas in this regard!



Best,



Sören



On 10.02.2020 05:59, Adam Sobieski wrote:

> Semantic Web Interest Group,

>

>

>

> I would like to broach, for discussion, scientific models and

> semantics, semantics in multi-model scenarios.

>

>

>

> There exist multiple models of atoms: the Dalton model, the Thomson

> model, the Lewis model, the Nagaoka model, the Rutherford model, the

> Bohr model, the Bohr-Sommerfeld model, the Gryziński model, the

> Schrodinger model, and the Dirac-Gordon model.

>

>

>

> It seems that scientific models can contain components which are

> symbols. It seems that language symbols, e.g. "electron" can be

> related to these model component symbols. It seems that these model

> component symbols can be related to abstract concepts, e.g. /the electron/.

> Perhaps, while the aforementioned models attempt to describe the same

> things, the overarching, more abstract, set of concepts which includes

> those described things, /the proton/, /the neutron/ and /the

> electron/, is itself, an abstract model.

>

>

>

> We can visualize a diagram, a graph, with a lexical symbol,

> "electron", on the left side, which is related, by arrows pointing to

> the right, to a number of model component symbols (Dalton_electron,

> Thomson_electron, ...). Each model component symbol is related to its

> containing model (Dalton_model, Thomson_model, ...). Then, as the set of

> models under discussion attempt to describe the same things, each

> model component symbol can be related to the same abstract concept,

> e.g. /the electron /(abstract_electron), on the right side of the

> graph, which can be from an abstract model (abstract_model).

> Furthermore, each model can be related to that abstract model. The

> lexeme "electron", from the left side of the visualized diagram, can

> also be related to the abstract concept, /the electron/, from the

> right side of the diagram, as it is another possible sense of the meaning of the lexeme.

>

>

>

> As an ideal natural language parser processed and interpreted the

> contents of a physics textbook, it might find that the lexeme "electron"

> meant different things in different chapters as the textbook's authors

> brought the audience on a journey through a number of models. The

> matter might become more pronounced as an ideal natural language

> parser or interpreter processed a set of physics textbooks, from

> kindergarten through university graduate level physics, and attempted

> to merge the contents together into a knowledgebase.

>

>

>

> I wonder what others in this mailing list might think about these

> topics (models and semantics, semantics in multi-model scenarios) and

> whether there might be any publications on these topics to recommend?

>

>

>

>

>

> Best regards,

>

> Adam Sobieski

>

>

>