4, language concept space statement primitive symbol system (second group MI)
At a demonstration of the 973 project, when I first announced 57 basic sentence clauses, Professor Ma Yude, a Professor of the School Chief Scientist, and the School of Science and Technology (the deputy director of the Ministry of Science and Technology), "I In your book, you are looking for these representations, that is, no more, it turns out to keep your confidentiality. "This is the professor of Ma Ma, he is not a linguist, but the image information processing expert, but he suddenly Seize the key to HNC.
"Semantic block is the function of the clause class - sentence classification - Basic sentence class 57- Hybrid class 3192"
This is the key point of the statement primitive symbolic system of language concept space
Where is the concept of a clause representing? This can be explained from multiple angles. Today I want to say: sentence class represents the semantic conversion of "The Lord Penny" concept.
Research on the concept of "The Subject" concept can be said to be the focus of linguistics in the language in nearly half a century, and there are many geography. The most famous of these most famous should have a conversion generating formula for Mr. Jimski
S = NP VP
This formula is not a semantic conversion, but it is the statement representation of a computer that can grasp, through it, you can grasp the concept of "the subject". In this sense, Mr. Jumski is the first pioneer from "people-oriented" to the "computer-oriented", and he is very savvy in strategy, first give an Indo-European system. "Robust" standard S = NP VP generic representation, and then sew it with the jurisdiction (GB) theory.
There are two representatives of the "Subject" Semantic Conversion Research. One is Mr. Phil Mo, which is the student of Josksky, and another is Mr. Shank, Shank, Jumsky. These two gentlemen realize that there is a function relationship between the subject (SVO), but unknown, avoiding the "functional relationship" proposal, using "grid" and "concept from the genus" terminology . why? Because if the proposed method of the function is directly used, it does not write a function representation, which will have an adverse impact on the paper as soon as possible.
Establishing a functional relationship representation is a joint pursuit of all scientific research, as well as all natural scientific branches from the initial signs from the initial omnipotent philosophy. Can a natural language statement creates some form of function representation? This is of course a very serious and significant topic. The "Ge" and "Concept" Theory I just tried to explore this topic, and the implicit assumption of the function from the subject and object is the meaning of this implicit. But this implicit assumption implies a significant mistake. Because there is three relationships between three mutual restrictions between the subject, it is not possible to act, and the two mutual restrictions are required, and the main belly does not care. The root of this mistake cannot be said to be the verb center saying (the statement of the Indo-European system is summarized) and the price of price grammar (another correct summary of the statement).
Considering that there is a relationship between the subjects between the three, it should be assumed that the three will be restricted by a more basic thing. If this thing exists, and name it is x, then you can follow the Joe Msky represents the following "Statement Block Representation":
J (x) = [jk (x)] [ek (x)] [{fk} |] (HNC20)
"The statement block representation" is the "Semantic Block Representation" of the "Semantic Block", its meaning is: The sentence j is composed of three types of statement units -jk, ek, fk, respective limited set, symbol "[]" indicates a limited collection . JK and EK are named as generalized object language volunteers and feature blocks, syndrome of the main language, which is generally equivalent to NP and VP. FK is named a complement block, which is equivalent to some PP, the parentheses {} of the fk indicates that it is not a must.
The statement block represents the fundamental difference between the HNC 20 and the Jumsky representation, not in replacing NP with JK, replacing VP with EK, and replaces the recursive of NP and VP with the respective limited sets of jk and ek (including empty set). HNC assumes that JK and EK are a function of a certain self-variable x. So, the problem is attributed to what this assumption is what is the X existing? The answer to the HNC is: sentence class. The literal explanation of the clause is: Statement concept Lenovo clipping type, also called statement level global concept Lenovo clipping, referred to as a clause.
So, how do you divide the sentence? Or what is the basis for the classification division?
The answer is clear. "The action effect chain is both the core concept of the core concept element classification, and is the master class classification." This means the highest level of statement.
The role effect chain is 6 links, then what is the saying of 7 basic sentences?
"7 basic sentences" still refers to the highest level of statement, 7 = 6 1, this " 1" comes to Aristotle's statement proposition says "Beijing is the capital of the People's Republic of China" Typical syndrome (basic judgment sentence) is not a category of the action effect chain. The proposition belongs to the judgment, "the effect of effect chain judgment" is also known as a broad sense of action, which does not have to be more important, but it is convenient to describe it.
57 groups of basic sentences are formal, then "57" this mysterious number comes from?
Mainly to the secondary concept node of the effect chain. The design process of these concept nodes should take into account the needs of basic sentence class, and the previous section has explained this. But it still feels not thorough enough, so it is here to refer to a paragraph in the "HNC theory", although this pass is about one of the difficulties, but there is only to abuse.
The design of the six main primitive primary concept secondary nodes must not only reflect the overall characteristics of each primary concept, but also a sub-class classification feature of each basic sentence class. This is the basic basis for the main primitive primary concept secondary node design.
The overall feature of the so-called primitive concept refers to content related to global association. If the action is to be associated with the ancestors; the effect is to be associated with changes and results; the process will be associated with the beginning and end, due to fruit, trend and transformation, and repetitiveness, etc. Sending, incoming, starting, starting, and end points, transfer content, tools, and ways.
However, not every overall feature has qualifications to form a clause class, and we only choose the feature of this qualification to serve as a secondary node, and put other features in the 0 branch. The content of the process (refers to the motion, evolution, and life process) at 0 branches, and the content (refers and information) constitutes the transferred secondary node, the reason is here.
"Hnc theory" pp85-86
So, "57" is equal to the sum of the number of the large-scale effect of the secondary node? no! "57" is the "group" number of basic sentence clauses, not "a" number. The division of "group" is based on many considerations. It should be noted that the basic sentence class represents 5 levels, but this level is equivalent to the professor of our country, not administrative cadres. It should also be noted that "57" is not can't change, and I even welcomes the genre of different numbers in HNC's own team.
4.1 Symbol System and Symbolization of Statement Blocks
The statement block indicates the physical significance of HNC 20, but such a representation is still in operation, and it is necessary to further symbolize the three semantic blocks in the representation. Symbolization First, you need to design the symbolic primitive of the main language block and their combination.
The main language is a function of the clause class, then how do I reflect this function relationship in the symbol representation of the main language? The most straightforward approach is: 1. Direct replacement of the symbol primitive of the spectrius block with the defined domain of the sentence class. Second, connect the symbolic primitive of the general object language block to the characteristic language symbol primary element. In this way, the semantic block is clearly expressed in the functional relationship of the sentence class. The definition domain of the sentence class is a generalized action effect chain. As the symbol primitive of the clause class, it is necessary to refer the corresponding symbol. HNC agreed: 7 links of the generalized effects chain with the following seven large English alphabetical representatives:
Eî {x, p, t, y, r, s, d}
E represents the general symbol of the sentence class, defines the specific sentence class primitive symbols in the domain, such as the process P, transfer T, the relationship R, the state S is taken from the first letter of the corresponding Word in English, but the action and effect is adopted. Special symbols x and y, this is due to the particular status of the two. Judging Judge's J, and Decide's D, that is the right, because J has been predetermined for code representation of sentence (juzi). It is undoubtedly unwise to use J, but I think it is necessary to distinguish between the sentence classification expression and the Jumsky's statement. At the same time, S is the natural choice of the status clause symbol, so The heart is coming, take a "big" in the world.
Now, it is assumed that the general symbol of the generalized object language block is G (which defines the domain as described below), then the statement block indicates that the HNC 20 can be rewritten into the HNC21 form that is easy to operate.
J (em) = [EMG | N] [EM] (HNC21)
Indicates that the digital subscript m in the formula HNC 21 indicates the level of the sentence class, and n is the relationship marker of the generalized object language volleyball. The accessory block is omitted in this representation, and it should be noted that G in the formula can be repeated.
Indicates that HNC1 and HNC21 represent all the results of HNC theory to today, and the surface wall is 6 years, and it is these two representations. The concept node table is evolved from HNC1, 57 groups of basic sentence clauses are evolved from HNC21. However, in the "HNC theory", he did not express these two representations, but also in individual places, deliberately got a little mystery or even misleading. This is because the market era is "fox" as the king's era, forcing honest "hedgehog", and not to stay a little bit.
Details indicating the type HNC21 illustrate more text. Only the agreement of the subscript M is described later. M The number of digits represents the level of the sentence class, one digit represents a first-class sentence class, and the two digits represent the secondary sentence class, and so on. The numbers of each "bit" represent the subclass type of the corresponding level sentence class. For example, a number of XM (m = 1-4) represents the four primary subclasses of the role sentence: the response of the action, the effect of the role, the role of exemption and constraint, one digit TM (m = 1- 4) 4 primary subclasses representing the transfer sentence, receiving, transfer, information transfer, exchange replacement and transformation; two-digit X20, X21, X22 represent general reactions, active reactions, and passive reactions React sentence (secondary subclass of the action sentence), T2B represents one of the self-transfer clauses of one of the secondary subclasses; the three-digit T492 represents the first exchange sentence; the four-digit T4A10 represents the extension replacement sentence. It should also be specifically described that the subscript M can be empty, representing the 0 (most advanced) sentence class, a total of 7, named the basic role in XJ, basic process PJ, basic effector yj, basic state SJ, Basic judgment JDJ, block expanded sentence DJ and action effect sentence XYJ. Talking about the level of the class is equivalent to our professor level, then the 7 0-level sentence class is very similar to the 4 special professors in my country in 1956. For G, the 2.2 session of "HNC Theory" "Thesis 2" is a non-leafless description, but unfortunately, the text is hard, and it is difficult to read. This power chart illustrates two main points: First, the semantic block must be handled as a function of the clause class, and the second is to introduce the concept of the keynote block of the key, and try to demonstrate that the primitive of the generalized object language is there and only 3: actors A, object B and content C. The introduction of the content element C is critical, so he does not hesitate to ink, which has made a near cumbersome argument from all angles. These three broad-listed object language volunteers are the defined domain of g.
Gî {a, b, c}
Letter A came to the Agent or Actor, b came to Object, C to Content. The repetition of G in HNC21 should be distinguished. In the semantic block representation of the sentence class, it means that ABC must be connected in sequence (mainly BC or AC connection), not reversed, and A, b, and C it cannot be repeated. However, when the phylogenesis is made, that is, the object of the phrase block is performed, and the content is decomposed, the three can not only reverse, but also itself can be repeated. E.g,
In this sentence, belongs to the concise state S041J = SB SC
SC = is still very good
SB = <"Professor Zhang" | Professor Zhang> Spirit
SBB SBC
SBBC SBBB
For SB, such as shown in two symbol lines, there is a decomposition of object content represented by the following equation.
SB = SBB SBC = SBBC SBBB SBC
The concise status sentence is one of the most common basic clauses in Chinese. There is no characteristic voice block, only two generalized object language volunteers. The sentence class knowledge of this sentence: SB may include SBC, but the SC does not include the SCB. Therefore, the boundary line of the two language volunteers is between "spirit" and "still", not between "Professor Zhang" and "Spirit". In 57 basic sentences, the boundary line between semantic blocks requires further discussion, and S041J is the only sentence class. Therefore, at least two papers at this seminar involve this class, I hope that the collisions can be benefited.
This section is to understand the key to HNC. As mentioned above, maybe more difficult, deeply, not fulfilled the promise of the loss. Now go to the next topic, hoping to help mitigate the understanding. 4.2 sentence classification and sentence class knowledge
Indicates that the HNC 21 is the basis of forming a clause class, but it is still not a clause representation. why? Because the clause representation must give the order in which the main language volunteers are arranged, the expression HNC 21 does not have a predetermined order. All representations given by the Size Table of the Basic Sentence Class are given the order of the main language, that is the clause class representation. For example, the basic functions mentioned above, basic procedural sentences, basic state sentences, physical transfer sentences, information transfer sentences, general response sentences, first out exchange sentence, extended replacement sentence, functional sentence, concise status clause They are:
XJ = a x b
PJ = PB P
SJ = SB S
T2J = TA T2 TB T2C
T3J = TA T3 TB T3C
X20J = x2b x20 XBC
T492J = TA T492 T4B2 T4C
T4A10J = T4B1 T4A10 T4B2 T4C
XYJ = a XY B YC, YC = (E) EC
S041J = SB SC
These basic sentence clauses represent the style of the HNC statement representation, of course, only part of the HNC statement representation, because there is also a representation of a hybrid sentence class and a composite sentence class.
Each sentence class represents each have specific sentence clauses, HNC hopes to put the so-called world knowledge as much as possible into sentence knowledge, which is sublimated to the concept level to express the world knowledge. Maybe it can be said that it can achieve this is whether the computer can understand the ultimate "bottleneck" of the natural language.
So, what basic content is included in this classic knowledge of mysterious "veil"? There is a discussion in the study, and Dr. Chuanjiang, the seminar, also has a special example.
Example 1: Basic Sewage Objects B, usually include 3 elements, HNC named them as a role object XB, effect object YB, and effect content Yc, so the general components of B semantic blocks can be written as the following representation formula:
B = XB YB YC
This expression is one of the sentence clause knowledge of the basic role, called a semantic block constitutes knowledge. For Chinese, this knowledge also has a very cute feature, that is, the order of the three elements must strictly abide by the representative indication, not reversing. In the P128 and P339 of "HNC theory", I will talk about this, please also refer to it.
Example 2: Basic procedural PB PB must be an abstract concept as a core element, and the basic state SJ's SB must be a core element in a concrete concept. Mr. Jumski is famous, but in fact, the sentence "colorless green thinking is fury", which should be attributed to the basic state of the sentence, but obviously does not comply with the sentence of this clause Class knowledge, thus it is very easy to determine.
Example 3: The transfer content T2c of the product transfer T2J must be in the specific concept, and the content T3c of the information transfer sentence T3J must be an abstract concept information as the core, and the T3C has block extended transparent, that is, it must be expanded to another statement, otherwise However, it can only be replaced with a referral phrase.
Example 4: The reactor of the general react sentence X20J must be a living body, first of all. The semantic block XBC representative of the reaction caused and its performance must simultaneously include specific and abstract concepts, frequent sentences. If the phrase block contains only specific or abstract concepts, it means omitting, and this omit is inevitably corresponds to a certain common sense knowledge. Example 5: First Excluding T492J and Extended Alternate T4A10J Each has three broad-scale object language volunteers, and the relationship between the three is described in the "Double Object", it is better to use "Double Subject" more appropriate because two classes The T4C has "Easy" phenomenon.
Example 6: Same as information transfers, with block expansion, belongs to the "double predicate" clause class. "Double predicate" means that two "central verbs" may be needed. For a large number of sentences in the language and block expansion, HNC believes that it is necessary to introduce the following two sets of characteristic voice block "symbol pairs" to describe:
(EG, EL) g = globle l = local
(EP, ER) P = premise r = result
Sentences and block expansion are compared in the Indo-European system, due to the influence of a central verb sentence specification, this hidden language is highly enlarged, and the processing of machine translation and Chinese verbs and the treatment of Chinese verbs are greatly beneficial.
Example 7: The basic state sentence is one of the clause classes without the characteristic language. The West is only used in the oral language, and Chinese uses a lot of speaking and written language.
As described above, it is only a scales of the scales of the sentence, but also a millennium of the ocean, but the current key is not quantified, but whether the knowledge of knowledge is easy to master or understand the computer. This shift in achieving knowledge form can be said to be the only goal of HNC theory is always pursued. The design of the "Language Concept Space Concept Biographic Symbol System" and "Language Concept Space State Symbol System" is surrounding this goal, the excavation and accumulation of sentence clauses is also around this goal.
Of course, there is a series of major theoretical issues that need to explore, including: assessment of clause knowledge potential, sentence knowledge and rules, relative and absolute knowledge of sentence knowledge, objectivity and impartiality of sentence knowledge. These texts will not be discussed.
(Fail)
