Re: [ontolog-forum] Representing design

[John Bottoms <john@xxxxxxxxxxxxxxxxxxxx>]

re: Representing Design...Using Reusable/Reproducible Research

Ciao Vostro,

I regret not being able to respond to your query earlier, I was working on a project. I have also been looking into research and design methodologies. I was a student of the late Italian designer Aldo Giorgini. I believe design and research share a number of components and both are intricate fields. Along these lines I have been looking into a recent development in this area, RR.

I think there are things in RR (Reusable Research) that could be interesting and useful. The discipline is still in its infancy but there are 1st generation systems in use. The concepts have been discussed for decades and some of the concerns have been addressed and implemented in existing tools.

"The RR goal is to make analytic data and code available so other may reproduce findings."

Basically the issue addressed is "How to write a research article for publication in such a way that it permits re-implementation of the original experiment(s)". Underlying this is the concern that as technology moves forward computers are inevitably involved. This means that the code, the machine running the code and the data form the foundation for a given research finding.

RR seeks to create systems in which the experiment components can be shared with the community. The completed published article contains all the ingredients to reproduce the research. This means having a document that includes 1.) narrative, 2) program code and data and 3) the meanings of the processing performed by the program.

In essence, all of these can be captured in a hypertext document. The current systems of XML/HTML do a fairly good job but there are some concerns going forward. Among these are extensibility, formats for multimedia and marking and security of components of parts of the document. The security issues are of concern to those of closed user groups with intellectual property or national secrets at risk.

The U.S. National Library of Medicine, NIH has published a paper on their work at this URL.
http://www.ncbi.nlm.nih.gov/pubmed/17911732

There are also sociological concerns about adoption of RR for larger communities. Jelena Kova�cevi´c's article "HOWTO ENCOURAGE AND PUBLISH REPRODUCIBLE RESEARCH" sheds some light on this issue.
http://jelena.ece.cmu.edu/repository/conferences/07_04_Icassp_Kovacevic.pdf

I ran into the RR field in association with biostatistics (one of the 'omics fields, such as "genomics") through the work done by Roger Peng at Johns Hopkins. His excellent paper can be found at AAAS free if you register on the website. A short article by him is at:
http://biostatistics.oxfordjournals.org/content/10/3/405.full.pdf+html

In the field of statistics, "Sweave" is the main tool for RR. Sweave uses LaTeX for graphics which makes it nice for writing equations. The website for Sweave is:
http://leisch.userweb.mwn.de/Sweave/#Sweave

I think we are looking forward to a more robust grammar for RR that can build on existing approaches. I am working on an article in this area but it will take me a couple of months to get it completed. There clearly needs to be an grammar or ontology for RR documents and IP.

I hope this helps.

-John Bottoms
 FirstStar Systems
 Concord, MA USA

On 10/15/2013 1:00 PM, Sandro Rama Fiorini wrote:

Dear all,

This list frequently rises the difficulties and ambiguities involved in modeling the notion of design (as in “product design”) and engineering model. We are now developing an ontology that must represent these concepts and we are facing those same issues. We would like to seek some advice from people in this list.

The ontology we are working on must deal with a conceptualization already in use by a specific application. This application deals with automated production lines that produce different types of artifacts (hammers, for example). These artifacts are produced according to their respective specific designs.

However, it is not clear what are these designs and what are their relationship with the respective artifacts. Intuitively, we consider artifacts as subtypes of physical objects. We consider designs to be mental or abstract entities that specify instances of an artifact to be built (but not *how* they are built, in terms of list of actions). For instance, a design of a type of hammer should specify what are its attributes, parts, structure and materials.

That definition of design makes it quite similar to the very notion of concept. The argument going on in our group is whether there is any essential difference at all between then, and if so, what are they relationship. Both seem to “abstract” the properties of a class of individuals. Differences could be the level of detail they are intended to specify. Another point of view emphasizes that there is a fundamental difference between the concept/class of Hammer (for example) and the design of specific hammers. Nevertheless, their main content of a class/concept seems to be very similar, but we are not sure whether this similarity necessarily implies the identity between these notions.

The differences become even more difficult to spot when we consider the design and its description. We distinguish the design from its description as a piece of text or a blueprint. A single design can have many individual descriptions.

One of the requirements of this project is that the description of a design must explicitly refer to entities provided by the ontology. For example, consider a design for a new kind of hammer, which should be 40 cm in length. This design has a description (say, in first order logic) “Ax NewHammer(x) -> length(x, 40cm).” According to our requirements, NewHammer and length must relate to the “Hammer” and the property “length” provided in our ontology. We could, for instance, rewrite that formula to “Ax NewHammer(x) -> Hammer(x) ^ length(x, 40cm)”. However, in this case we start to have the description of a design of Hammer is striking similar to the description of a subconcept of Hammer.

Some people in our group quite understandably have problems with that position, arguing that the content of the description of a design should only reflect an existing concept, but it cannot be the same as the description of a concept (to justify the subsumption relation in the second formula, for instance). Other argue that the description of a design *is* the description of a concept, justified by the lack of differences between them.

The question is: what is the relation between design and concept/class? We considered two possible answers:

1-The  concept/class Hammer’ is its own design. In this sense, it seems that the notion of design is only useful as a modeling tool to group different possible descriptions of the same entity. It seems to be interesting to think about the instances here. In this point of view, both the concept and design of NewHammer collapse in the same entity, abstracting the same class of individuals.

2- The concept/class NewHammer and its design are different things. An instance of NewHammer is different from an instance of Design that specifies the class of NewHammer. In this point of view, it seems that we are proposing a new type of relationship that holds between instances and classes, which is not the instantiation relation. In this relationship (maybe called specifies), a specific instance (of design) specifies a new class of entities in the ontology. That is, this new class comes to existence due to the very existence of an instance of Design that specifies it.

best,

-- 
Sandro Rama Fiorini
Phd Candidate
Institute of Informatics
Federal University of Rio Grande do Sul (UFRGS)
Website: www.inf.ufrgs.br/~srfiorini

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)