Research or Practice?

As a research computer scientist, it has been a struggle for me to understand when my work is research and when it is not.  Recently, the UTEP Office of Research and Sponsored Projects (ORSP) decided to enforce a new compliance module for research guidelines.  It has similarities to what we use to do and I already took the previous course, however, I was not required to read the entire Belmont Report 1979, released by the National Commission for the Protection of Human Subjects in Biomedical and Behavioral Research.

This time, I did and there it was, the clear understanding ... I think :-).  The issue is interpretation but for me the meaning was clear.  Section Part A states:
" 'practice' refers to interventions that are designed solely to enhance the well-being of an individual patient or client and that have a reasonable expectation of success. The purpose of medical or behavioral practice is to provide diagnosis, preventive treatment or therapy to particular individuals ... 'research' designates an activity designed to test an hypothesis, permit conclusions to be drawn, and thereby to develop or contribute to generalizable knowledge. "

My research is based on the premise if we share related research information on the Semantic Web, we make the research more accessible and understandable overall.  My original IRB was focused on research teams and their data when all along I was interested in how to reason over collections of related research information.  However, I needed to support scientists in capturing their research and sharing it on the Web, before I could consider the benefits of the Semantic Web.  The result was that I was not understanding how to characterize the support of scientists in sharing information in terms of my research.  Now, I believe I get it.  The support for scientists, data management, was a 'practice' that was to enhance the capabilities that scientists have in describing their research and sharing it on the Web.  This is something scientists do everyday and the support I gave was to help them share their research in a team-focused Web portal.  Our research center talks about this a lot, in particular when researchers leave due to graduation or when research efforts finish.  No real research there ... I had always called this a 'means to an end'.  Nevertheless, I was unclear whether I needed to include that human-involvement as part of my IRB.  Consequently, I did.  The main drawback was that I was having to consider how I would manage the team and privacy, etc. as part of my research and I was struggling with how I was going to get this type of buy-in when my research was trying to expose Semantic Web benefits to get people interested in the first place.  Basically, a cart and horse dilemma.

Now, with my clearer understanding, my IRB amendment is based on the methodology that transforms the information for access on the Semantic Web and understanding the benefits to leveraging Semantic Web techniques in sharing scientific information.  The practice of helping scientists share on the Web is not my research.  The practice of understanding what it means to transform this information on the Semantic Web, to improve accessibility and understandability, is.

... so, that is my enlightenment for today.

Are Research Objects an extension or omission of the Semantic Web (Linked Data) vision?

I read the following paper ... something I had planned on reading for a while.

http://semanticommunity.info/@api/deki/files/4703/=research-objects-escience.pdf

Some of the authors are well known in the Semantic Web community, but especially, they are well known in the collaborative executable workflow environment (my experiment).  What I am struggling with is their position - because this appears to be a position paper.  I feel like they are saying that Linked Data, a quality of the Semantic Web, is missing descriptions of a resource object : the aggregation of resources, their relationships and structure. 

Basically, in the Background section's last paragraph of page 2 and the 3rd paragraph of page 3, they seem to state that Linked Data is missing 'the Semantic Web' where things are self-describing (things can be collections) including structure and relationships (expressed with the languages to describe things).  Linked Data may not specify that but the Semantic Web does enable what they claim is missing (and again, Linked Data is part of the vision of the Semantic Web) - it's like saying that pie making is missing from baking ... but baking enables you to make pies? If that is the case, everything is missing from Linked Data and the Semantic Web.  Even their discussions in section 4 seem to delve more into the details of how to implement the Semantic Web, not extensions to it. 

What I thought they were going to say was that Linked Data does not support actionable behavior (links to executable process) which, I suppose it does not explicitly, it supports self-description which 'could' mean actionable behavior ... that would explain their argument for research objects ... something they introduced in their my experiment project.  In the end though, they focus more on e-laboratory infrastructure (its characteristics and principles) which aligns with their research objects and the myexperiment work.  They even start to get into documenting a scientific investigation which is more aligned with my work; but in the end they focus on the results of some execution of an experiment - still rather singular.

I obviously do not have the insight they do so concluding this definitely makes me uncomfortable.  This work does relate to mine except 1) I am not specifically considering actionable behavior to reproduce results of an experiment, I am considering research collections documenting the many results, experiments, publications, etc. of a research effort. where workflows with actionable behavior could be included and 2) I believe the basic Semantic Web gives me pretty much what I need to describe research, aside from user understanding and more time :-) .  I may find deficiencies in the modeling languages but that exposes issues with the implementation, not the Semantic Web vision.

Thanks to some great mentoring, my dissertation is forming nicely.  Things had gone awry for a while ... not the CI-Server, just the research direction and mentoring.  I am happy that my ideas and interests have come full circle, they are forming the foundation for what I had proposed upon beginning the CI-Server framework.  So...back to progress.

I have been able to abstract the methodology of my research from the framework.  Probably should have been the other way around, either way, this week's post will focus on reason-able views.  These are discussed by Ontotext, the makers of OWLIM-Lite and OWLIM-Full.  There are similarities to how they are managing conceptual spaces and there are two papers out where this has been implemented by two research efforts.  The biggest difference I see, just superficially, is the way we manage the triplestores.  Development-wise, it seems like we had the same ideas around the same time; they use their triplestore and I use ARC2.  I expect programmatic and easy creation and re-creation of the triplestores, it seems like their implementation is less flexible in that sense.  Their entire implementation, however, is certainly more developed than mine - although, I am sure the ideas will be complementary ... we are documenting collections of inter-related semantically structured information and storing them in a single data repository.  Certainly, we can both gain from seeing how to manage the conceptualization space this way.

CI Server updates soon

SSDBM paper accepted.  This was a good foundation for collecting data.  I have ported that in the new CI Server client library.  This will be available soon but I don't know if I will have the panels available by then.   http://rio.cs.utep.edu/ciserver/CI-Downloads will have it.  This will require me documenting a lot of features including the test tool, config tool and desktop.  All have features that interact with a server.  I expect to have CyberShare, Rio and Trust servers available soon as well but this immediate version will not have all the group support we need.  I had to rewrite some code for the general functions and to provide some basic menus.  Having these different CI Servers interact with the CI Client tools will be possible with the new server.  Overall I wanted to understand what features needed to be at the server and which at the client.  I think the server side is more understandable now.  The suggestive query tool will help me understand how to use the project based data on the server to help understand research.  Next: Knowledge Annotation panels in API,  Group support in server, RDF views based on user requests - simple searches.

CI-Server Model

So this has taken me a while.  I want to keep the culture of a site separate from its data management.  Drupal has really been great in helping me understand this.  Now, to show all node metadata as rdf and providing some simple exposure calls and I should have a good handle on this.  This is outside the RDF API realm because I don't want to have to map my types.  I just want them exposed as RDF not just forms. 

Doing this and I would be able to show that I can manage such data and provide a simple regurgitated web mine ... no additional knowledge yet.

My next step would be to document it.  SSDBM might have been a good conference but too late.

... off to a meeting.

SIOC xml

Well, after I looked into it, it turns out that the SIOC xml model will satisfy what I need.  It can be difficult to map the content and choose containers but it worked for the Kepler integration.  So, for now I can collect the comments and publish them as SIOC xml comments about documents that are being discussed by scientists.  Further annotations can be made for uri's in a semantically annotated document.  My next step is understanding how to evaluate the content and make suggestions.

Comparing Social Sites

Starting to model discussions - in particular comparing social networking sites. Twitter, Facebook ... I'll determine others as I get there. There is so much variety but I believe that there are enough common characteristics that I can do this