酷兔英语
    当前位置:
  1. 首页
  2. 文章
  3. 英语能力
  4. 英语词汇
  5. 分类词汇
  6. 正文

自然语言处理词典(The Natural Language Processing Dictionary)3

0 收藏 0 分享 0 赞

章节正文

D

 

 

DE
= discourse entity
DE list
See history list.
declarative
= indicative.
demonstrative
A kind of determiner, that is, an ingredient of noun phrases. This class of words includes "this", "that", "these", and "those". They are part of the referencesystem of English. That is, they is used to tell which of a number of possibilities for the interpretation of the rest of the noun phrase is in fact intended. Demonstratives are most useful in spoken language, and are often accompanied by a pointing gesture.
derivation
A derivation of a sentence of a grammar is, in effect, a proof that the sentence can be derived from the start symbol of the grammar using the grammar rules and a rewriting process. For example, given the grammar 1.S → NP VP, 2. NP → ART N, 3. VP → V, and lexical rules 4. ART → "the", 5. N → "cat", and 6. V → "miaowed", we can derive the sentence "the cat miaowed" as follows:
S⇒NP VPrule 1
 ⇒ART N VPrule 2
 ⇒the N VPrule 4
 ⇒the cat VPrule 5
 ⇒the cat Vrule 3
 ⇒the cat miaowedrule 6
One can then write S ⇒* "the cat miaowed": i.e. ⇒* is the symbol for the derivation relation. The symbol ⇒ is referred to as direct derivation. A sentential form is any string that can be derived (in the sense defined above) from the start symbol S.

 

descriptive grammar
The sense in which the term grammar is primarily used in Natural Language Processing. A grammar is a formalism for describing the syntax of a language. Contrast prescriptive grammar.
determiner
Determiners are one of the ingredients of noun phrases. Along with cardinals and ordinals, they make up the set of specifiers, which assist in reference - that is, determining exactly which of several possible alternative objects in the world is referred to by a noun phrase. They come in several varieties - articles, demonstratives, possessives, and quantifying determiners.

 

See also here.

discourse entity
A discourse entity (DE) is a something mentioned in a sentence that could act as a possible antecedent for an anaphoric reference, e.g. noun phrases, verb phrases and sentences. For example, with the sentence "Jack lost his wallet in his car", the DEs would include representations of "Jack" "his wallet", "his car", "lost his wallet in his car" and the whole sentence. The whole sentence could serve as the antecedent for "it" in a follow-up sentence like "He couldn't understand it" (while "Jack" would be the antecedent of "He").

Sometimes discourse entities have a more complex relation to the text. For example, in "Three boys each bought a pizza", clearly "Three boys" gives rise to a DE that is a set of three objects of type boy (B1: |B1| = 3 and B1 subset_of {x|Boy(x)}), but "a pizza", in this context, gives rise to a representation of a set P1 of three pizzas (whereas in the usual case "a pizza" would give rise to a DE representing a single pizza.)
P1 = {p | pizza(p) and exists(b) : Boy(b) and y = pizza_bought_by(b)}.
The function "pizza_bought_by" is the Skolem function referred to in lectures as "sk4".

discourse entity list
See history list.
distinguished non-terminal
See context-free grammar.
ditransitive
A verb in English that can take two objects, like give, as in "He gave his mother a bunch of flowers". Here "his mother" is the indirect object and "a bunch of flowers" is the direct object. The same sentence can also be expressed as "He gave a bunch of flowers to his mother", with the direct and indirect objects in the opposite order, and the indirect object marked by the preposition "to". The preposition in such cases is usually "to", or "for" (as in "He bought his mother a bunch of flowers" = "He bought a bunch of flowers for his mother."

Bitransitive verbs can appear with just one or even no syntactic objects ("I gave two dollars", "I gave at the office") - their distinguishing characteristic is that they can have two objects, unlike intransitive and transitive verbs.

Here is an incomplete list of ditransitive verbs in English.

 

E

 

 

ellipsis, elliptical
Ellipsis refers to situations in which sentences are abbreviated by leaving out parts of them that are to be understood from the context. For example, if someone asks "What is your name?" and the reply is "John Smith" then this can be viewed as an elliptical form of the full sentence "My name is John Smith".

Ellipsis causes problems for NLP since it is necessary to infer the rest of the sentence from the context.

"ellipsis" is also the name of the symbol "..." used when something is omitted from a piece of text, as in "Parts of speech include nouns, verbs, adjectives, adverbs, determiners, ... - the list goes on and on."

"elliptical" is the adjectival form of "ellipsis".

embedded sentence
An embedded sentence is a sentence that is contained inside another sentence. Some examples, with the embedded sentence in italics:

 

  • John believes that Mary likes pizza
  • If Mary likes pizza then she may come to our pizza party.
  • If Joan liked pizzathen she would come to our pizza party.
evoke
A noun phrase is said to evoke a discourse entity if the noun phrase refers to something related to a previously mentioned discourse entity (but not to an already-mentioned DE). For example, in "Jack lost his wallet in his car. Later he found it under the front seat.", the phrase "the front seat" evokes a discourse entity that has not actually been mentioned, but which is in a sense already present as part of the the DE created by the phrase "his car".

See also anaphor.

exists
"exists" is a textual way of writing the existential quantifier, which is otherwise written as an back-to-front capital E. It corresponds fairly closely to the English word "some". Thus,

 

exists(X, likes(X, spinach))

would be read as "for some entity X, X likes spinach" or just "something likes spinach". This might be too broad a statement, as it could be satisfied, for example, by a snail X that liked spinach. It is common therefore to restrict the proposition to something like:

 

exists(X, is_person(X) and likes(X, spinach))

i.e. "Some person likes icecream." That is, we are restricting the type of X to persons. In some cases, it is more reasonable to abbreviate the type restriction as follows:

 

exists(X : person, likes(X, spinach))

See also forall, Skolem functions and this riddle.

experiencer
Experiencer is a case that usually fills a similar syntactic role to the agent but where the entity involved cannot be said to act. It is thus associated with the use of particular verbs like "remember", as in "Jim remembered his homework when he got to school". Here "Jim" is the experiencer of the "remember" situation.

 

F

 

 

failure of substitutivity
In some situations, things that are equal cannot be substituted for each other in logical forms. Consider believe(sue1, happy1(jack1)) - jack1 may = john22 (i.e. the individual known as Jack may also be called John, e.g. by other people, but Sue believes John is happy may not be true, e.g. because Sue may not know that jack1 = john22. Thus john22 cannot be substituted for jack1, even though they are equal in some sense. See also Allen pp. 237-238.
features in NLP
Features can be thought of as slots in a lexicon entry or in structures used to build a logical form. They record syntactic or semantic information about the word or phrase. Examples include the agragreement feature, the sem feature that records the logical form of a word or phrase, and the var feature that records the variable used to name the referent of a phrase in a logical form.
first person
One of the choices for the person feature. A sentence is "in the first person" if the subject of the sentence is the speaker, or the speaker and some other individual(s), as in "I like pizza" and "We like pizza".

"I" and "we" are first-person pronouns, as are "me", "us". Other words with the first-person feature include "mine", "my", "myself", "ours", "our", and "ourselves".

FOPC
This stands for First Order Predicate Calculus, a standard formulation of logic that has logical operators like and, or, and not, predicate symbols and constants and functions, and terms built from these, together with the quantifiers "forall and exists. It is common for semantic representation systems in NLP to be expressed in languages that resemble or are based on FOPC, though sometimes they add significant features of more elaboratelogical systems.
forall
"forall" is a textual way of writing the universal quantifier, which is otherwise written as an upside-down capital A. It corresponds fairly closely to the English words "each" and "every". Thus,

 

forall(X, likes(X, icecream))

would be read as "for every entity X, X likes icecream" or just "everything likes icecream". This would be too broad a statement, as it would allege that, for example, rocks like icecream. It is usual therefore to restrict the proposition to something like:

 

forall(X, is_person(X) ⇒ likes(X, icecream))

i.e. "Every person likes icecream." That is, we are restricting the type of X to persons. In some cases, it is more reasonable to abbreviate the type restriction as follows:

 

forall(X : person, likes(X, icecream))

See also exists.

free morpheme
A free morpheme is a basic or root form of a word, to which can be attached bound morphemes that modify the meaning. For example, "happy" is a free morpheme, which becomes "unhappy" when the prefix "un-", a bound morpheme, is attached.
FUT
See modal operators - tense and tense - future.
future perfect
See tense.


文章标签:词典  

章节正文
Copyright © 2023 Qeto All rights reserved.
商务合作 | 联系我们 | 关于我们 | 版权声明 | 隐私条款 | 如果智培

京ICP备05059198号-1 京公网安备 11011202003401号