Commonsense-feasibility scheme

REPRESENTATION OF THE COMMONSENSICAL FEASIBILITY (OR OTHERWISE) OF A PROPOSITION

We are encountered with statements like these in research literature on Commonsense AI - 

Televisions don't eat ice creams.

You cannot make salad out of shirts.

You cannot eat noodles on a globe of earth.

In most data, 2 entities/concepts come together. (TVs - ice creams; salad - shirt; noodles - globe of earth.) 

Here is an algorithmic method to represent, decide & explain the acceptability or inacceptability of a phenomenon (like those above) on the basis of commonsense : 

(The method will become clearer with the 3 examples that follow it.)

Let the 2 entities be A and B.

1. Think of the most basic definitive property of A and that of B. Let them be A1 and B1 respectively.

2. Check if A1 is compatible with B1.

            If not, drop the issue. 

3. If compatible, move to A2 and B2. A2 and B2 are the next most definitively basic properties that one can think about the entities A and B.

Check if A2 is compatible with B2. But that not enough; here is a point - 

A2 is not a “mathematical” evoluent of A1, or B2 is of B1. So the earlier lower levels also have to be checked with for compatibility. That is, alongwith A2-B2, A1-B2 and B1-A2 compatibilities also have to be checked.

And so on. The moment there is incompatibility, the issue drops dead and the phenomenon is unfeasible on the basis of commonsense. 

The more commonsensically invalid the phenomenon/issue will be, the earlier will it drop dead; and vice versa.

Now, let's see this applied to 3 cases - both commonsensically feasible (the first 2) as well as otherwise (the 3rd one).

1. TVs cannot eat ice-creams.

We have TV & ICE-CREAM as A and B.

Which is the most basic definitive property of a TV? One might say that the first thing that comes to mind is that it displays moving pictures. No, even more fundamental to that is that it's firstly, a thing - a non-living thing. When you first see an off-TV, that's the first thing you consume about it. 

So, A1 = non-living thing

Similarly, B1 = non-living thing

Checking the compatibility of A1 and B1 :

Can a non-living thing eat a non……..here itself there is a basic infeasibility that a non-living thing cannot eat anything. So A1-B1 is incompatible and the issue drops dead here itself. 

The very first link ‘A1-B1’ is broken and there is no solid link.

So at the very first level, there is incompatibility; hence the ridiculousness of the proposition.

2. You cannot make salad out of shirts.

We have  A = SALAD & B = SHIRT.

THe most basic definitive property of a salad is that it's a thing - a non-living thing. 

So, A1=non-living thing.

Similarly, B1 = non-living thing.

A non-living thing can be made out of a non-living thing. So there is compatibility between A1 and B1. 

Now, move to A2 and B2 : 

A2 = it is something edible; B2=made of cloth

Can something made of cloth be edible? NO. So A2 is not compatible with B2; and the issue drops dead here.

We might also check for A1-B2 : Can a non-living thing be made of cloth? YES.

And we might also check for B1-A2 : Can a non-living thing be edible? Depends upon what sense you take the non-livingness of the eatable as. So YES or NO depending upon that.

So, we have solid A1-B1, A1-B2 and B1-A2 links, but a broken A2-B2 one.

The issue collapses at the second stage itself; hence the ridiculousness of it.

3. You can use a mug to carry water

A = MUG, B = carrying water

A1 = non-living thing

B1 = solid, supporting carrier

A non-living thing can be a solid, supporting carrier. So, A1 - B1 is compatible.

Lets move to A2 and B2 : 

A2 = something with space to carry

B2 = should be sufficiently rigid for carrying and not spill-worthy (tearable/too thin/loose)

Something with space to carry can be sufficiently rigid for carrying and not spill-worthy (tearable/too thin/loose). So A2-B2 is compatible.

A1-B2:

A non-living thing can be sufficiently rigid for carrying and not spill-worthy (tearable/too thin/loose)

So, A1-B2 is compatible.

B1-A2 : 

Some solid supporting carrier can have space to carry.

So, B1-A2 is compatible.

Then we move to A3-B3 :

…...and so on.

So now we have A1-B1, A2-B2 as well as the cross-links A1-B2 and B1-A2, all compatible. So it's a nice strong robust structure capable of withstanding reality. Hence it is a sensible proposition.