Example-Based, Science-Based, and Rule-Based Modeling

> As the human development necessarily started with "learning by examples" (=EBM), the historic development (from Arabia\ Greece over Galilei and Leonardo up to Newton and Hawkins opened a cosmos of science and Sciencs-Based Modeling (SBM), which practically enabled and heavily influenced today's human life and society. 

> Unfortunately, EBM mostly got stuck with the restricted capabilities of human brains, who cannot handle more than a maximum of 3 to 5 parameters. SBM however experienced another gigantic enhancement by the development and rise of computers, which led also to some over-done SBMs, like eg calculating the stiffness of a car body by a more than 500.000 parameter FE-model, or employing today's biggest computers 24\7 in weather prediction, just because "we have enough computing power". So, practically, Modeling is SBM only in these days.

> Besides, there are several powerful statistical models in use . But every statistical model in it's kernell is based on some kind of assumptions (= prejudices), which potentially may lead to unknown errors of the results, especially in complex multi-parametric situations. Moreover, statistical models, if reliable and helpful, easily can feed their results as examples into EBMs. So, practically, they can be seen as parts of EBM world.

> The same goes for another important know-how source, eg in product and process design: Testing. It delivers several examples, more or less fitting eg the design goals, and so often more or less valued by the test owner. But EBM can use all of the results, especially the bad ones, to bring new insights and new solutions to the problem at hand.

> Last not least, Rule-Based Models (RBMs} can play an important role in KHE projects. Not so much in the shape of Expert-Systems anymore, like proposed some decades ago, but often enough as arbitrary restrictions in a parameter space or restrictive relationships between some problem parameters. But as such, they most often can be easily integrated and fulfilled in an EBM's parameter set and KHB, respectively.

> In this light, EBMs or better networks of EBMs can be seen and used as a very powerful environment to collect, consolidate, integrate, concentrate, USE, and permanently refine the know-how from arbitrary sources (eg SBM, test, RBM, experts, etc pp) in any problem domain and application area, no matter how complex, interdisciplinary, or (un-)explored the problem at hand may be.

> As an example, the whole development (and manufacturing) know-how of a car (as a complex today's product) can be collected, stored, integrated, and refined in a meaningful structured network of EBMs, containing the complete and refined experience and know-how of a big team, gained over the years. Not only the know-how of a single car, but of an extended range of car types, the car company might be interested in. So, the "development of a new car" can be reduced to a work of minutes (compared to some years today), by employing the appropriate network of EBMs, and something like "customer-specific" car design (and manufacturing) can become reality.

> So it's about time to remember EBM, and to emphasize and extend the restricted EBM capabilities of human brains by developing and deploying reliable, problem-oriented networks of EBMs, which can concentrate the know-how from any sources, and can represent and handle today's complex inter-disciplinary problems in a convenient and powerful way.

> Along this way, opportunities can be opened to use the plethoria of examples available in our world, but treated and wasted most often as "trash" up to now. Everything = Examples! Any source can be integrated! Let us start to (re-)use the know-how implicitly available in examples!. So everything done and happening can be understood in this way as (a sequence of) examples to feed appropriate EBMs, to gain and to re-use the know-how contained, and to identify new know-how, hidden in the plethoria of examples available.