It is known that the PhD would make an original contribution and/or advance knowledge in a given field. In my field ( Computer science), there are various PhD contribution that combine methods to solve a problem.

In short, my question is: What are the conditions to develop an original/novel method for a contribution of a PhD by combining two or more methods that had never been combined before. I mean the case where we combine method A and method B to solve a problem T where A and B has never been combined before to solve T.


One way to decide if you have something novel and important is to ask whether the combination is more than the sum of its parts.

To take an important example from mathematics, Newton and Leibniz aren't credited as the creators of Calculus because they (independently) invented the derivative and the integral. In fact those had been known for a long time. Integration by infinitesimals had been known for centuries as had differentiation and continuity, though for a shorter period.

What made the studies of Newton and Leibniz (most) important was actually the Fundamental Theorem which combined the two and showed their relationship as inverse processes.

Something similar happens in medicine, though in a less fundamental way, when a new use is discovered for an old medicine, such as was the case with aspirin and heart disease.

There is no reason that the same doesn't apply in CS. How is the combination more than just the sum?

| improve this answer | |

Same as any kind of research. First have a state of the art of existing methods, which will describe and analyse what is currently done. You should point out the benefit and shortcomings of these methods, and highlight how these single methods do not fit the purpose of the problem. The search should be as exhaustive as possible within the bounds of he problem.

You can then proceed to explain your solution, and its result on the given problem. Analyse its value and shortcomings as you did previously.

When you will have done all this, you will basically have proved that:

  • No existing method currently provide the necessary solution for the problem at hand (the combination of methods aren't already tested anywhere either, and all single solutions fail the benchmark of the experiment).
  • The combination considered is solving the problem (or is not solving the problem either, with different outcomes), is correctly bench-marked and understood.

Possibly, you might have to justify that the combination of the two methods aren't an obvious combination that would be used by any non-expert in the field having some sensible basis in the problem being solved.

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.