An Expedient Mind: ENTITY WORDS

AN EXPEDIENT MIND: ENTITY WORDS

Tartan Hen Publications : Home | more books | Contact : feedback@tartanhen.co.uk

ENTITY CONCEPTS (Nouns and noun phrases)

In this chapter I will describe the structures which represent the meanings of various common words. As I mentioned earlier, the correspondence between the common grammatical classification of "noun" and what I call "entity concepts" is not perfect, but it is close. "Entity concept" is a semantic classification. An entity word is one which refers to some kind of physical object, like a stone, or a person. An entity has a location and exists over a period of time. As a grammatical classification "noun" indicates the relationship between words. But it also has a semantic dimension. It includes things of the physical entity kind but it also includes abstract concepts like "justice" or "kindness". I will deal with abstract concepts elsewhere. For the present I will deal with simple physical objects only, and to make things as easy as possible I will deal with only solid physical objects which have a definite shape and size.

{JOHN}
The simplest kind of entity word, is one which refers to a unique object. In other words, it is the name of an object. So this is an #entity with a specific ID. If we think in terms of the inheritance hierarchy of #entities, this is one at the bottom of the heap. It has no subordinate #entities.
NOTE: Recall inheritance hierarchies
It inherits most of its properties from the #entity {MALE-PERSON} which will, in turn, inherit properties from {PERSON} and it will get properties from {PHYSICAL-OBJECT}. We can assume, therefore, that {JOHN} will have all the normal physical properties and body-parts we associate with a person. If, for some reason, {JOHN} does not have the standard properties of a {PERSON} then these special properties will be added to the {JOHN} concept. (Having only one leg for example). The concept #{MALE-PERSON} confronts us with a few additional considerations. Just what is the difference between a male person and a female person? The topic is fraught with sociological as well as computational difficulties and it is the case that what distinguishes men from women varies a great deal, as these terms are understood by a child and by an adult. For a child in pre-puberty and before the facts of life are explained by an adult, male-female differences will be confined to sociological stereotypes. Our robot, however, is not going to grow up and mature in that way. We willneed to decide at the outset what we are going to tell him and I see no way we can shrink from explaining the full anatomical and biological basis of sex. More #scenarios - sexual intercourse - ecstatic pleasure - conception and the eventual birth of a child.
There are other components to this male-female thing. As experts in the art of commercial advertising know only too well, certain images, shapes, scents, tactile sensations, are triggers for pleasurable sexual arousal. The trouble is, these triggers are different for men and women. So if we want our robot to behave outwardly like a human we will need to decide at the outset whether he is a male or a female robot. Up to this point I have referred to him as “he”, and so I will continue to assume that he has a male psyche, even if he does not have the full anatomical attributes of a male person. He will find the curvacious shape of the female body a definite “turn-on” (a chi-experience) and the sight of a naked male either slightly nauseating or faintly absurd (and definitely a psi-experience).
“John” is an identifier for a particular person. It is entirely possible, however, that there will be several different “John”s within the experience of any person or our robot. In these circumstances the use of a cache to define the context of discourse (see the chapter:- INTERPRETATION) will usually remove ambiguity. Failing that, it will be open to the robot to ask the question - “which John are you talking about?”

{A}
In my description of words above I used the expression “a stone” “a cube” and “a river”. I avoided the use of the somewhat more enigmatic expressions “stone”, “cube” and “river”, and I did so for good reason. The word “stone” does not refer to {A STONE}. It refer to the substance stone and to a generalised type, or class of entity, not to an actual entity with its own identity. To turn such an expression into a reference to an actual entity with its own identity we must use the phrase “a stone”. So the word “A” does a clever trick. It takes the concept {STONE} as a template and creates a new individual #entity {A STONE} which has all the properties defined by {STONE} and in addition it has a unique #entity identifier of its own. It adds the ID to {STONE}.
Recall the way we represent a multiple object. It is represented by a dynamic list which has a generator function and a template. What we have now is an infinite set of all stones. The template is {STONE} and {A} is the generator. It creates yet another element {A STONE} in that very long list of all possible stones.

{THE}
This word does exactly the same thing as {A} except that it also suggests that the #entity in question has been mentioned before. That triggers a search through previous bits of the #interpretation (including the cache store) to try to find that previous mention. Another side effect of {THE} is that the #entity in question is established within the context of discourse and is temporarily established as the focus of attention.

{STONES}
The use of a plural word like “stones” is represented by a dynamic list. As such, it has a generator and a template and therefore does not require the addition of a word like “the” or “a”.

{LARGE}
As in “The large stone”. Traditionally classified as an adjective, the word “large” has as its meaning the #concept {LARGE}. This #concept is a modifier. It carries within it a reference to an anonymous #entity. This #entity carries very little information and no individual identity. But it does have a size which is larger than the standard size for such a #entity (whatever that is). It is the task of the interpretation procedure to find that #entity and to apply the additional information about its size being larger than standard (for such ‘entities). In this way a large mouse is identified as a mouse which is larger than the standard or expected size of mice and a small elephant is identified as one which is smaller than most elephants. There is no implication that the large mouse might be larger than the small elephant which would be the case of we used absolute definitions of large and small. Noun Phrases and the Scope of Adjectives The phrase “A large stone” is classified in conventional grammar, as a noun phrase. According my analysis, the meaning of a noun phrase is a structure which represents an #entity. The noun-word “stone” supplies the template for the properties of the #entity, the adjective-word “large” supplies additional information about one or more of those properties, and the determiner “a” establishes the unique identity of the #entity. A noun-phrase, by itself, merely states that a particular #entity is present within a hypothetical environment which constitutes the field of discourse. A sentence does more. It normally provides information about some action or state of affairs, within the hypothetical environment. The #entity, represented by the noun-phrase, is then available to fulfil a role within that action or state of affairs. There is, however, a small technical difficulty associated with the analysis of a noun-phrase. It concerns the problem of identifying which word within the phrase is the template for the #entity and which merely supply additional information. This problem arises when there is more than one adjective or when a word which is normally considered to be a noun, is used in an adjectival role. Consider this phrase “An advanced passenger train” It is not obvious whether the #concept {ADVANCED} should supply extra information to the template {PASSENGER} or whether both {ADVANCED} and {PASSENGER} supply extra information to the template {TRAIN}. The #concepts {PASSENGER} and {TRAIN} will each contain a #scenario which represents the transportation of people in a vehicle of some kind. {TRAIN} will be specific about the vehicle and will identify that vehicle as the salient item. {PASSENGER} will not be specific on the issue of the vehicle but it will identify the people within it as the salient. It seems that it is not particularly important how we assign roles to {PASSENGER} and {TRAIN} in the manner of conventional grammar. The result of putting these two #concepts together, will be the creation of a template which represents a railway vehicle with people on board. The problem, which does remain, however, is the ambiguous role of {ADVANCED}. Does this tell us about the age of the passengers or the technical design of the train? The two interpretations are, I feel, equally valid. Had I never at any time heard that phrase before, I would be quite uncertain about the correct interpretation. The #concept {ADVANCED} will contain two typical use #scenarios. One will refer to technical design and the other to advanced age. If one is given priority over the other, in terms of the order in which they are listed, then the interpretation will be slightly biased in favour of the first in the list. The same effect can be seen with the phrase “a geriatric passenger train”. The word “geriatric” normally refers to the age and infirmity of people. But it can be, and is sometimes used to describe venerable equipment in a deplorable state of repair. Finally, if the robot finds it impossible to disambiguate a phrase, he can ask the speaker which interpretation is correct. Since human beings have similar difficulties, and frequently solve those difficulties in the same way, a failure of the system to assign a definite interpretation to such a phrase, should be considered a strength and not a weakness.

Tartan Hen Publications : Home | more books | Contact : feedback@tartanhen.co.uk