Week5: Data Exchange & Interoperability–R&R Jinxuan Ma

Both the Schaefer and Thanos articles make it clear that interoperability still has a long way to go but they remind me that a great deal of progress has been made since the 1990s when I used MacLInkPlus to convert documents written in the Macintosh version of WordPerfect to PC’s running MS Word. Data translation software has come a long way from those days and the advent of open standards has made a major difference in the ability of programs to share information. AnIML is interesting because it is potentially creating a data structure that will allow a complex process to be modeled using a simple text based editor and the tagging conventions of its particular markup language. And as the author of the article notes, “While not necessarily convenient, this feature is critical for long-term data retention scenarios: even if we lose access to the software, we keep access to our data” (Schaefer, 14). As part of my article reading for the week, I came across an interesting article with implications for AnIML. The article, “Illinois researchers build Dropbox-like storage, analytical system for scientific data” describes framework called 4CeeD (http://t2c2.csl.illionois.edu (Links to an external site.)Links to an external site.) can be linked to scientific instruments and allow scientists to upload their data to cloud architecture. While the article doesn’t mention AnIML, it would seem to be a good tool to use in conjunction with the cloud architecture. Klara Nahrstedt, the principal investigator of the project, reports they “have developed a cloud architecture that makes it easy for scientists to not only upload their data but also curate and manage the data, as well as get real-time search results.”

The ideas of mediation and interoperability run throughout the articles by Thanos and Barnaghi et al. and point to the need for various forms of “scaffolding” to allow the system to mediate between various descriptions of data and the encoding systems that convert data into discrete objects that can be processed by digital computers. Barnaghi tackles the complexity of modeling the world through discrete methods of encoding. The graphic on page 10 provides a nice encapsulation of the issues and many levels of descriptive encoding that are necessary for the continuous flow (i.e. analog) nature of physical reality to be “chunked” into discrete units of coding with various levels of representational meaning. This is essentially what James J. Gibson recognized as early at the 1960s was problematic with cognitive models of intelligence. The over-optimistic forecasts of computer scientists and AI researchers in the 50 and 60s has meant that scientists are having to come to grips with a different paradigm of information transmission that is more similar to biological models as Gibson argued for in both 1966’s The Senses Considered as Perceptual Systems and 1979’s The Ecological Approach to Visual Perception. Thanos is still largely arguing for a logical and discrete model of computing when he argues for various methodologies for creating interoperability and a rather top-down approach to tackling the problem. I would counter that nature evolved many systems of intelligence that mediate between multiple data streams continuously and simultaneously and I would suggest that while the various suggestions made I the article are sound, they are limited by a failure to examine the problem from the perspective of evolved intelligence. Barnaghi at least acknowledges the need to capture data from sensors and other transducing systems.

Mediation introduces the need for informatics to wrestle with semiotics, especially the systems of both logical and visual analysis that were pioneered by the American polymath Charles Saunders Peirce in the 1800s. It seems self-evident that computer scientists and informatics practitioners would benefit by reading more deeply in the existing literature of other fields rather than stumbling through a mostly computer science derived approach to issues. And vice versa, discoveries being made in informatics have direct relevance to ecological psychology and semiotics.

The takeaway from this week's readings are: (1) that mediation and interoperability is a hard engineering problem, but a necessary one if there is to be continued progress in machine learning and improvements in removing processing bottlenecks from informatics systems.

(2) Researchers in the field should go beyond they BOK of their particular disciplines and look for existing solutions or problem sets that have already been tackled by other thinkers in other disciplines. Progress is retarded by the parochial approach to problem-discovery and solutions.

Coming from a strong background in philosophy and evolutionary biology, I am constantly struck by the parallels and overlaps between information science topics and areas of research in the aforementioned fields. Cybernetics began its academic life as a topic in biology in the works of Jacob von Uexkull in the 1910s and 1920s. Mediation finds its roots in the work of the philosophical study of being and epistemology. So, I find those informatics topics resonate strongly with my existing knowledge base. And I find myself increasingly experiencing déjà vu as I read articles for the class that feels like mirror-images of topics I’ve thought about in different contexts. Informatics, as I've said before, is mostly the engineering side of the mirror and that’s reflected in language and argumentation constructs it uses.

Barnaghi, P., Sheth, A., & Henson, C. (2013). From data to actionable knowledge: Big data challenges in the web of things. IEEE Intelligent Systems, 28(6). 6-11.

Schaefer, B. (2013, December). A fresh look at the AnIML data standard: Data standards promise easier collaboration, data exchange and interoperability in the informatics world. Scientific Computing,13(3). 13-15.

Thanos, C. (2014). Mediation: The technological foundation of modern science. Data Science, 13, 88-105.

University of Illinois News Bureau. (June 19, 2017). Illinois Researchers Build Dropbox-Like Storage, Analytical System for Scientific Data. Retrieved from https://csl.illinois.edu/news/illinois-researchers-build-dropbox-storage-analytical-system-scientific-data.