Science & Semantic Technology

Of all the areas where Semantic Technology may help to transform current practices, no one area may be impacted more than Science.

I’ll distinguish empirical science from the myriad of other sciences by stating that it is characterized more by processes designed to facilitate discovery – the scientific method. The goal of empirical science is to solve problems, it does so through answering a series of questions, often through use of experimentation. Of the IT domains I’ve discussed previously the one that is most involved in pure science is Healthcare, so let’s take a look at that for moment.

In Medical Science, we base clinical practice to some extent on the results of treatment focused experiments or experiments which seek to isolate a particular causal connection to a medical condition. Using the current incarnation of the scientific method, a researcher begins with a hypothesis – let’s say that a particular gene predisposes a subset of the population to contracting a particular disease. The experiment then seeks a demographic group that seems more prone to catching the disease in question and then tests for the gene that had been included in the hypothesis. The experiment either confirms or denies the original premise / idea / hypothesis – often times this requires a recursive return to the experiment with slightly modified variables in order to work through the discovery process.

This is what we are familiar with – the way it is done now. It isn’t too much different from how we’ve been doing empirical research for nearly 200 years with the exception of the surrounding technologies involved. The one thing that has always stuck me about it though is that in a great majority of cases we are looking to discover something we already think we know – it is often more about validation than true discovery.
Problem Spaces & Problem Solving: Problem Solving begins through Semantic Characterization
In other words, if we truly don’t understand a problem yet – how effective is it going to be to try to prove our hunches based on our previous biases? Worse yet, if our experiment can only effectively manage the inter-relationship of two or three key variables in a situation which may have hundreds worthy of consideration, how can we be totally certain that we’re not telling ourselves what we want to hear instead of truly understanding the complex interactions that cause a particular problem.

So, we are limited in Science today by three things:

1 – A discovery method which to some extent favors validation and problem simplification.

2 – The lack of necessary technology to facilitate a more open-ended discovery approach and the ability to model and deconflict data generated by an ever growing set of experimental variables.

3 – The lack of a modified methodology that allows us to consider more complex test scenarios and provide a management approach for the technical elements needed to facilitate complex experiments and test result analysis.

Enter Semantic Technology.

One way that I think to think of Semantic technology is as context management – whether this relates to specific software tools that claim to do this isn’t as important as the notion that we are now developing the capability to manage issues and problems within the full extent of their relevant contexts. The immediate challenge with application of this philosophy and its incumbent technology is that we (collectively) will begin to understand that there are more gray areas than we thought – that the either / or answers we built for ourselves from traditional experimental controls are not so clear as we’d like them to be. Once upon a time some folks used to refer to this as Fuzzy Logic and perhaps others tried to characterize through Chaos Theory but the bottom line is that it is simply a recognition of real world scenarios.

In the real world, issues are almost never neatly contained within easily manageable and interpretable bubbles. If we want to make the innovative leaps necessary to move beyond where we are now one of the most important things we will need to do is evolve our assumptions in regards to how Science is performed. Once we begin to do that it will be become ever-clearer how important Semantic Technology will become for Science.

Semantic Technology will be the driving factor for: 

  • Real-time & Near-time Heuristic Collaboration.
  • All forms of Artificial Intelligence – previous attempts to create AI have fallen short precisely because the notion of what intelligence is in the context of computational behavior has been somewhat misunderstood. At this juncture we don’t so much need a pure reasoning capability, what we really need is the ability for knowledge relationships to be imprinted somewhere and be easily retrieved by all (based upon our preferences and expectations). From this foundation more can come, but the knowledge web / semantic web needs to be in place first – it provides a sort of global intelligence building block for all other future applications to exploit. (in other words why teach each AI entity to learn all the necessary knowledge themselves individually – why not focus attention on how to exploit the knowledge web)?
  • Discovery Validation
  • Curriculum Development & Curriculum Mapping (within the context of collective and personal learning environments)
  • Unstructured Discovery (i.e. open-ended problem-based research that begins with flexible sets of possible Hypotheses).
  • Comprehensive Empiricism – in other words, studies done utilizing all identifiable variables associated with a given real world problem.

These and many other applications will combine to allow for more complex experimentation and more rapid community assessment of emerging concepts. Semantic technology will be the glue that binds global knowledgebases and global problem solving communities. It will allow us and Science to become more dynamic than ever been before.

Let’s look at an example from Healthcare IT – one of the most important elements of clinical practice is the ability to provide consistent process approaches for specific treatments. Some folks refer to these as decision trees or treatment algorithms, others just see it in the form of flowcharts or diagrams that are shared to ensure that caregivers follow certain steps in given situations. Once applied within any given provider location, the results of a new practice can be assessed against previous approaches or even assessed against the absence of an approach. A decision tree, flow chart or process – however we wish to refer to it, can be captured using semantic formats. The practice is captured, let’s say as a concept map that is transferable elsewhere in OWL. This artifact then represents knowledge that can be shared to any other healthcare provider on the planet. The processes and their related resulting clinical data can be managed by communities – allowing for improvement, comparison and of course discovery of related techniques or concepts. As the communities develop, the ability to assimilate new or improved clinical practice will increase exponentially both in terms of scale and speed.

One takeaway to consider for today – often new technologies are considered “disruptive.” This carries with it a distinct negative connotation and that reputation comes from the experience of many previous tech hype cycles – a poor reputation often well deserved. I contend that Semantic Technology is on the level of the concept of structured databases (DBMSs) as providing a transformational capability or perhaps even more important than that. It is revolutionary rather than disruptive in that it solves many of the problems related to all the previous hype cycles by allowing us to place all of those technologies in context – for the first time ever. Semantic Tech provides context management on quite a few levels…

Copyright 2010 – Stephen Lahanas