Vocabulary Notes

A programmer submitted some vocabulary, asking for critique. This note shares corrections in vocabulary/concept creation, highlighting vocabulary creation ideas by example.

    define	battle-n1
		label		battles
		label		battle
		child-of	war-n1

… this needs a plabel for the plural.  Ask yourself the question, “is a battle a war?”  The answer is probably “no”, ergo, battle is not a child of war.  

As an example of a relationship that is valid, consider “banana” and “fruit”.  Is a banana a fruit?  Are bananas orthogonal to apples, pears and grapes over the concept of fruit?  Yes, so fruit would be a good candidate for orthogonality, too.

Anyway, back to battles… what is the parent of a battle?  It’s an activity.  It is associated with war.  It is orthogonal to preparation, retreat, victory... Maybe it is orthogonal to other “war activities?”

define  war-n1
  	  	label		wars
  	 	label		war
  	  	child-of	hostility-n1
  	 	orthogonal	hostility-n1

Is war a hostility?  I’d agree to that.  What other kinds of hostilities are there?  Embargo, sanctions, etc…  They may be considered orthogonal to each other.  

But, is peace also orthogonal to war?  Yes.  Is it an hostility?  No.  It’s definitely not a child of hostility.  What is it, then?

define  country-n1
   		label		country
		label		nation    
   		plabel		countries
   		plabel		nations
   		child-of	place-n1

This might be good.  It’s defined already, though.  Do you have the existing definition?

define  	mexico-n1
    		label		mexico
    		child-of	country-n1
    		orthogonal	country-n1

 This looks right, though I think it is already defined.

define  frog-n1
		label		frog
		plabel		frogs
		child-of	amphibian-n1
		orthogonal	amphibian-n1

Okay


define	toad-n1
		label		toad
		plabel		toads
		child-of	amphibian-n1
		orthogonal	amphibian-n1

Okay


define	amphibian-n1
		label		amphibian
		plabel		amphibians

This might work.  However, ‘amphibian-n1’ is not the child of anything else.  But, amphibian-1 has to be the child of something ultimately leading to ‘things.’  Otherwise, consider that I might ask ‘is a frog an amphibian?’ and get the answer ‘yes’.  But, if I ask ‘is a frog a thing?’, the answer would come back as ‘maybe.’

Moreover, there are/may be many inferences related to ‘things’ that should apply to frogs, such as ‘if a thing is near another thing then the latter thing is near the former thing.’  I would want to be able to say ‘the frog is near the toad’ and have ‘the toad is near the frog’ computed by inference. For that both have to be ‘things’.

define	forest-n1
		label		forest
		plabel		forests
		child-of	place
		orthogonal	tree

Not sure why ‘tree’ seems like a reasonable orthogonality.  Often, concepts are orthogonal to other concepts with the same parent(s), so if forest were a child of ‘place-n1’, then forest might also be orthogonal over ‘place-n1’.  That would make more sense.

define	desert-n1
		label		desert	
		plabel		deserts
		child-of	place

Looks pretty okay.  However, it is not orthogonal to any other kinds of places, so a person could be ‘in the desert’ and ‘in the ocean’ at the same time.

define	cactus-n1
		label		cactus
		plabel		cacti
		child-of	desert-n1
		orthogonal	plant-n1

Conceptually, you might think that a cactus is a child of the desert because that’s where you find cacti.  However, when you ask yourself ‘is a cactus a desert’, the answer is ‘no.’ In Brainhat-land, the answer will be ‘maybe’, unless cacti and deserts are also declared to be orthogonal.

Digression: why, when you ask ‘is a cactus a desert?’,   would come back ‘maybe’? The reason is that during processing, Brainhat could learn from you or through inference that a cactus is a desert. To illustrate how: do you remember the Simpson’s episode where Homer wants to be employee of the year but Mr. Burn’s gives the award to an inanimate carbon rod?  So, learning about this story we might ask “is a carbon rod an employee?”  Normally, no.  But it may become one during processing, even if it seems ridiculous. 

    
    define	rocket-n1
		label		rocket
		plabel		rockets
		child-of	things-n1

This is valid, but it puts rockets too close to the top of the taxonomy—you know, ‘animal, vegetable, mineral or rocket?’

 define	castle-n1
		label		castle
		plabel		castles
		child-of	place-n1

This one is syntactically correct, but if a country were to be a place (as in a definition farther above), then a castle is kind of on the same abstraction level as Mexico.  As long as castles and Mexico aren’t orthogonal, this will function.  But, a castle should probable be orthogonal to other kinds of buildings—schools, restaurants, houses(?).  
  

    
    define	lantern-n1
		label		lantern
		plabel		lanterns
		child-of	things-n1

As with the definition of rockets above, this is awfully close to the top of the taxonomy.  It would work, but it would be resistant to distinctions between rockets and lanterns and animals.

    
    define	harbor-n1
		label		harbor
		child-of	place-n1
		orthogonal	ocean-n1

Ask the question: “Is a harbor an ocean?”  Nah.

    define	highway-n1
		label		highway
		orthogonal	road-n1

A highway needs to be the child of some thing(s) that lead ultimately up to ‘things’.

    define	valley-n1
		label		valley
		plabel		valleys
		child-of	terrain-n1
		orthogonal	mountain-n1

‘Terrain’ seems like a good choice for a parent.  It would probably be a reasonable choice for orthogonality, too.  In this definition, we have ‘valley’ being orthogonal to ‘mountain.’  But, a valley is never going to be the child of a ‘mountain.’  In processing, springs might shoot out.

    
    define	house-n1
		label		house
		plabel		houses
		child-of	place-n1

‘place’ might be better substituted by ‘building’, or something.

    define	island-n1
		label		island
		plabel		islands
		child-of	terrain-n1
		orthogonal	mountain-n1

Like ‘valley’, above, ‘island’ could be considered a terrain.  So, it would be better if it were orthogonal to something like ‘terrain.’

    define	town-n1
		label		town
		plabel		towns
		child-of	place-n1
		orthogonal	city-n1

“Is a town a city?”  I think the answer is “no.”  So, this looks correct.  However… see next definition comments.


define	city-n1
		label		city
		plabel		cities
		child-of	place-n1
		orthogonal	town-n1

You can use any concept you like as the anchor for orthogonality.  Making ‘city-n1’ orthogonal to ‘town-n1’, and vice versa, might work technically.  But then you would have two anchors for orthogonality—‘city-n1’ and ‘town-n1’.  If, then, you subsequently wanted to define hamlet or village, you would need to make these orthogonal to both ‘city-n1’ and ‘town-n1’.  This would be ugly and inefficient.  It is preffered to have a common orthogonal concept for all of them.  

Note 1: you do not want to use the concept that you are defining as the anchor for orthogonality.  Say, for instance, that you define ‘city-n1’ and make it orthogonal to ‘city-n1’.  You will have made a concept orthogonal to itself.  This is actually done in processing to eliminate concepts.  In this case, you wouldn’t be able to use the concept ‘city-n1’.  

Note 2: You can define a concept as an anchor for orthogonality that needn’t have any parents; it might exists simply for the purpose of distinguishing concepts from each other.  For example, this is legitimate:


define	city-n1
		label		city
		plabel		cities
		child-of	municipality-n1
		orthogonal	geographic-location

define	geographic-location