3, language concept space concept element symbol system (first group "m i")
The study of various mathematical spaces has formed an important idea of the spatial primitive. If a space finds a complete set of primitives, the characteristics of this space can be accurately expressed. Clearing the idea of the primitives into language research is the RCSchank of the United States. Because the gentleman is too "out of the rebellion", there is nothing famous in the Chinese language circles, but in the referential references, it is special. His five works were selected.
Mr. Shanke made a "hedge" in-depth study of "hedgehog", think it is one of the concepts of language concept space. However, how many language concepts are comparable to the concept of "transfer"? Mr. Shanke has taken "fox" mode.
HNC's conceptual base set of language concepts continues to conduct a "hedgehog" study, and "transfer" "transfers" "compatriots sisters" are found, they are: role, process, transfer, effect, relationship and status. Transfer is just a member of 6 "Sisters". They make the core space of the language concept space, named the main base concept, also called the effect chain.
With regard to the effect of the effect chain, there is a paragraph, because it is often quoted, it has become a "set", which is not repeated here.
The center of "Penny" means that the 6 links of the effect chain are 6 basic sides of anything. If the 6 sides of a thing have been fully described, the face and characteristics of this thing are clear enough. The so-called knowledge of a thing is that the final analysis is the expression of these 6 sides, the so-called understanding of a thing is to grasp the information and knowledge of these 6 sides. Since the statement is the expression of things, the effect of the effect is of course the core content of the statement. Therefore, the acting effect chain is both a class of core concept primitive classification, and a division of the statement classification. This will mention this next section.
The effect chain is just a class of concept primitives in language concept space, then how much is this "class"? Many pioneers explore this vital issue, most explorers consciously unconsciously embarked on the idea of simulating biological taxonomy. However, language is not a biological, far more complicated than macroscopic biology, only induction, analysis, and synthesis methods are not enough, and also need to fusion interpretation and hypothesis test. The objective language concept space in the brain, the knowledge of cognitive science and brain science can provide, but leaving the grand goal of revealing the mystery of the brain, but only the first step in the long language. The more realistic exploration ideas may wish to rely more than a point of interpretation and assumption.
Based on this idea, HNC assumes: language concept space first can be divided into specific and abstract two major sub-space (or two regions). The specific concept refers to the concepts that can directly correspond to specific objects, such as "Mountain, Lake, River, Sea, Plant, Animals, People" "Country, City" "Factory, Store", etc. Abstract concept refers to the concept that cannot be directly related to specific things, or can only be related to the properties of a particular thing, which is two subclasses, collectively referred to as abstract concepts. The former, such as "role, process, effect, relationship, state", etc., the latter, such as "concept, thinking, emotion, consciousness, morality, advocation, punishment, corruption, quantity, quality, nouns, prefix", etc.
3.1 4 assumptions and 4 categories of spaces
The first subclass of the abstract concept is the effect chain, as illustrated above, which makes a core space of the language concept space, which is the first assumption of the language concept space. Such concepts have a distinct feature that language philosophy "can refer to" the concept "concept is not completely applicable, because it is" nothing "and" ignorant ".
The second subclass of abstract concept is a bit "more variable" means, but human activities, including psychological activities and thinking activities, obviously "a large piece", which is naturally the main object and content of the language. According to this, the second assumption can be made directly around the outer concept space around the core space, named composite primitive concept space, referred to as the composite primitive concept. With the body's primitive concept space and composite primitive concept space, the "缥 多 多" abstract language concept space is not so "embarrassing". We collect these two types of concept space syndrome. The basic feature of this space is: there is a clear core and a huge outer layer of the same content.
Although the primitive concept space is huge, but can not contain all the abstraction concepts, what is the abstract concept collection of big pieces? Some basic objects of philosophy and natural science together can make another sub-space in abstract concept space, and named basic concept space, in fact, this is the third assumption of language concept space.
After the primary concept space and basic concept space, the "缥缥" abstract language concept space has been quite avatar. Now I should consider the complete problem, think about any obvious vulnerability. The vulnerability is clear, that is, the language concept space and the natural language space transition or mapping are required. This needs can be said to be the "tool" required for conversion, which is the abstract concept contained in the narrow shape and generalized morphology, including the so-called virtual word of particularly rich in Chinese. HNC named this concept as language logic concept, which is a fourth assumption that the language concept space is composed.
With 4 abstract concept subspace based on four assumptions, the abstract "area" of the language concept space is complete? This topic issue for mathematicians is needed to adopt the "superloy" attitude that is mentioned in the quotient. Now, the key to the problem is not the complete mathematical proof, but 4 assumptions. But before testing, a specific design is given to the math structure of the 4 sub-spaces.
3.2 Abstract language concept space number symbol design
The abstract language concept space symbol design is essentially the "reputation of the natural language symbol system", the design object is the concept of language concept space, and the natural language symbol is mainly vocabulary. So this design can also be seen as a red design of natural language vocabulary symbols. However, its implementation process is to summarize in natural language space, and then in language concept space, different than WordNet or "How to" is only summarized in natural language space.
The term concept of the primary primary system seems to be a little mysterious, but it is actually a thin layer of window paper. "Mathematical structure", a string of digital symbols. However, relative to natural language symbols, this digital string must be changed in the following 3 points: First, each digit of the concept of the primary primitive number string has determined and unique meaning, and the natural language (especially pinyin language) The syllable string or alphanumeric string, its single syllable or letters are generally not determined, and the whole of the strings has certain significance, and most of them do not have uniqueness. Second, the requirements of the concept base number string start from the starting point at any point, and there is a corresponding overall meaning, and the words of the natural language generally do not have this feature. Third, the three basic features of the concept primitive, that is, three basic contents, hierarchical, internal relevance, external relevance, and appropriate sorting of digital strings, and naturally Language symbols cannot have this expression.
Based on these three requirements, the digital string of the concept primitive must be designed.
Y | (M | T |) |
General form, where the symbol "|" indicates that the variables there are repeatable variables. Here the digital string Y | indicates the level of the concept, referred to as high level; digital string M | represents the internal correlation of the concept, referred to as the middle layer; digital string T | represents the concept of external relevance, also called network, referred to. Symbol (M | T |) | Indicates (M | T |) as a whole and repeatable. The M | or T | may be empty set, which means that the order of M | and T | can be exchanged, that is, the high layer can be directly entered into the bottom layer, and the middle layer can occur after the underlayer, but the high layer is always at the foremost. Y | (M | T |) | The specific implementation of the means can have two options, one is to add a marker for the middle layer and the underlying symbol, the other is a non-marking, only the high-level digital string is agreed. Number, and give different numbers in the middle and underlying. The HNC symbol system selection is a method of representation. Some typical examples are given below, and then the overall design of the symbols will be further described. In order to facilitate the readers who are not familiar with HNC, the symbols () and [] of the marker are each differentiated, respectively, respectively, respectively, respectively, respectively, with high-level symbols.
HNC Concept Node Sample Table
High-level representation
0 role
00 "Exemption" and "constraint"
01 assignment
02 Life response to the role
03 exemption
04 constraints to make the object "do not"
1 process
10 basic characteristics and basic types
11 process
12 process cause source flow
13 process trends and transformation
14 new metabolism and life and death
2 transfer
20 basic characteristics of transfer
21 receive
22-product transfer
23 information transfer
24 exchange, alternative and transformation
5 state
54 structure
54-body structure
54-0 surface structure
54-00 line structure
54-000 point structure
7 psychological activities and mental state
71 psychological activity
711 attitude
7115 Attitude in Interpersonal Communication
High school underlayer
00 [8] Physical role
00 [9] chemical effect
00 [A] Biological
10 [b] life process
10 [b] (c5n) (n = 1 young, n = 2 less, n = 3 cyan, n = 4, n = 5 old)
11 (E5N) (n = 1 start, N = 2 end, n = 3 continues)
22B self-transfer
23 [9] information orientation information
23 [9] (1) Question
23 [9] (1) [9] Question
23 [9] (2) Answer
23 [9] (2) [9]
23 [9] (EA4) Relying on the directional information of a certain relationship (recommended)
23 [9] (EA5) Since the top (indication, command, approval)
23 [9] (EA6) from bottom to (report, please report, report)
23 [9] (EA7) There is no lower level relationship, strong expectation (requirement)
23 [98] statement
23 [99] Goodwishes beneficial directional information (persuasion, criticism, warning) 23 [9A] malicious harmful directional information (accusation, 诽, intimidation, deception)
23 [9b] Response to malicious harmful orientation information (辩, refute)
7115 [9] Communication Attitude
7115 [9] (E41) is not humble
7115 [9] (E42) Humble (might)
7115 [9] (E43) (arrogant)
These examples embody Y | (M | T |) | all feature of the structural formula, the number uses 16-based.
The hierarchical level of the concept is not difficult to get a relatively clear impression through the "7-71-711-7115" concept sequence. The hierarchical features include semantic top and lower, and the upper and lower relationships are usually used in the Kui Lan Dynasty, which is the expression of the language space. The reader may wish to compare the concept space here, and make your own judgment.
The internal relevance of the concept includes the negative, contrast, including three aspects, the corresponding middle layer symbols are
Dual EMN; N n = 0-3; 4-7
Contrast CMN; DMN
Contains -; -0; -00
The negative concept is an extension of the concept of antonym concepts, and the contrast concept is the symbolization of the concept of the concept. As can be seen from the examples, the concept of duality is a concept that needs to be discussed in depth, not the "opposite unity" "Converse Unity" invented by the Great Philosophy "Hedgehog" Can be summarized, and there is also a discussion in this seminar.
The concept of the concept is very complicated, and the underlying symbol of the external correlation is indicated by digital symbol 8-b. However, in fact, each underlying symbol can be expressed in a combination of high middle layer symbols, which means that the underlying symbol is substantially high-level symbolicization or simplification. This resolved process may involve "consciousness, re-recognition, memories" and core mystery, aunt, do not dare to talk. From a practical point of view, the setting of the underlying symbol simplifies the calculation of conceptual correlation, which is the basis or power of HNC still working hard to design the underlying symbol. "However, the underlying design is a complex system engineering, we hope to cooperate with linguists and peers." The appeal of this paragraph of the university is still effective.
3.3 Subspace design of language concept space
The design of the language concept space is actually the division or design of the concept category. This issue has been described in the preface and 3.1 of this section. Here, the following two points: First, the overall description of the language concept space and the full symbol of the concept primitive; the second is the main embodiment of the interpretation process in language concept space design.
3.3.1 Space Description of Language Concept Space and Concept Biography
"Wonderful is not as good as seen", a virtual color picture of course is the best show for the overall language of language concept space. Unfortunately, the old man will not use the old set, as shown in the following table.
Language concept space
Abstract concept space concrete concept space
Main complex base syndrome P w group
Physical in this words
Summary Overview of Based Logistics
Yuan Yuan Mun Editor's Concept
Overview Overview
Niannianniannian
J L JL S f, H, Q x P, PE, W, PW, JW
2 3-4 2 2 2 3 Hanging on the hooking 26 8 9 12 2 4 7
The points of this table are as follows:
First point, language concept space can be divided into abstract concept space and concrete concept space, abstract concept space points 7 submospheres, concrete concept space points 3 sub-spaces, there is a transition or two-canable physical concept sub-space. The letter line in the table marks the marker of each sub-space, also called the concept type symbol, the main body, and the composite primitive concept, with the Greek letter φ for shared type symbols, has been discarded. The grammar concept sets 3 symbols, but only f in digital symbol strings. Both Class P and W class only list two types, not full. The first line of numbers in the table represents the number of high-level bits of the corresponding concept, only "psychological response and mental state" in the composite primitive concept, and others are 3. The meaning of "hanging" is that it does not have a numerical symbol string, relying on the connection with the abstract concept symbol, such as W54-is structural, W54-0 is a surface structure, PW22B is the transportation tool, p10bc55 is old people. The second line of the table represents the number of root nodes of the corresponding subasive space.
Second, in addition to the syntax concept in the abstract concept, there is a five-way group characteristic, in which the five-way group of the primitive concept is particularly complete, five yuan groups and various combinations are called concept categories. The complete expression of the concept of the concept is:
[Type Sign] [Category Symbol] [Nutrie String] (HNC1)
The symbol indicated by the formula HNC1 is named HNC mapping symbol, and the semantics of language vocabulary can be expressed by HNC1 and its combination. In this way, the semantic expression is converted from the natural language space to the language concept space, and "symbolic arbitrary" to "symbolic correlation" conversion provides a computed symbolic basis for computer semantics.
The meaning of the five-way group is explained in detail in "HNC theory", and it is also a special article with the etiological relationship this seminar. Here is only a little, that is, the argument of Chinese mean, if you put it in the language concept space, maybe more easy to clarify your ideas. Mr. Li Jinxi's argument on Chinese "Words Nothing, Class Answorked", Mr. Gao Kai is a unique argument on the word problem. Now it seems that if Fan is in the primary concept space, then it should be said to Li, high two Mr. Mr. is unpredictable, but cannot be extended to all language concept space. The narrowness and general meaning of the form is the meaning of "of", "or" so ". The abstract concept has the five-way group characteristics, and Chinese is due to mono feature and corresponding square words. There is only a countermeasure for the fifth group characteristics, so there is a morphous and praise, and the rich form in dialect does not change Chinese. One fundature, why can't you reach a consensus?
Third, the type of language concept space can be regarded as the type of semantic field, and the language concept collection of each root node within each type is a specific "semantic field". The field has a type, different types of fields have different characteristics, and research should be studied. The awareness of physics, Mr. Einstein has no results for several decades, and the "unified field" of linguistics talks. However, it is possible to study various specific semantic fields, and the HNC concept primitive symbol system is to carry out this study, providing a different space with simple natural language space.
In the fourth point, the HNC concept primitive symbol system is high, the medium-level node is a nor, and each underlying node can be seen as a compound enrich, the complete problem of the set of suits is once daunting, with HNC The establishment of the concept base symbol system, although the completeness of the nor is not proven, it is already possible to adopt an "supernatural" attitude. Semantic works often said: "Semantic field analysis and vegetarian analysis have proposed some instead of all semantic analysis, only for limited semantic space, which is still unable to compete for all words." Now, this statement needs to be modified . 3.3.2 Interpretation of the language concept space design
The determination of the root node of each sub-spatial node of the language concept space is mainly an induction process. This induction process is a process of co-equivalent to the commonality and personality of the words until the highest level is reached. This step-by-step hierarchical "processing" process is of course not a relaxed thing. Fortunately, only 1,200 commonly used Chinese characters provide unparalleled convenience conditions for this "processing" process. These semantics are fully enacted by Chinese characters that form tens of thousands of back-to-directional and forward-connected double words in modern Chinese (both constitute "orthogonal" vector), which is contained in these double words. Concept of Lenovo information, its own classification, clarity of the context, is called the "wonders" of language information resources. From this "watching", the maximum commonality of "role, process, transfer, effect, relationship, and status", is not a matter of hard work. Therefore, the "HNC theory" said: "Here, the author cannot express his respect for the ancestors of the creation of Chinese characters. It can be envissed that if Phil Mo and Mr. Shank are rut to Chinese, the concept hierarchical network theory may appear 20 years ago. "
After obtaining the root node of each sub-space, the high-level design of each root node is mainly deducted. Taking the root node "action" as an example, the closest concept of the role is to bear the body, because if there is no toilet, the role is "empty", meaningless "effect", it is not necessary to describe it. That is to say, the role must be accompanied by withstand, "The Action Belt" must be a "role" root node. After the auspress is affected, it will inevitably produce a certain effect. If the toe is a living body, this effect is specially named "reaction". Life is not negcasably in response to the role, so "the reaction of life" must also be a "role". From the role itself, there are two special forms of role necessary to consider, one is to cancel or exempt the role of a role, and the other is to generate a constraint. Why do you want to think about it? Because the statement expressing these two effects has a special statement knowledge that is different from the general effect, it is specialty in the content of the object (equivalent to the syntax), that is, the next section to explain Sentence knowledge. In this way, the root node needs to be given in the "HNC Concept Node Example Table", also known as the secondary node.
The interpretation process of the so-called high-level node design is the above two aspects: First, the concept of root concept is derived, similar to the biological children said by the saying go. The second is some special sides of the root concept itself, expressing these sides of statements contain some special statement knowledge. These two main lines have a generality, or that it is assumed that they are the concept of all root nodes Lenovo main line, then the thinking process along these two main lines is interpretation, not inductive.
The high-level design of the process and metastasis has also vividly expressed the above interpretation process. "The process of the process" "The cause source of the process" "process" process "process" is "process" biological children, and "new metabolic and life and death" is a special side of the "process". Similarly, "Receive" is a "transfer" biological child, and "product transfer" "information transfer" "exchange, alternative and transformation" is a special side of "transfer". There are two interesting phenomena here that it is worth noting that "the process" has 3 kids, and "transfer" only "reception"; the other is 10 defined as "the basic characteristics and basic type", and 20 is defined as "Basic Features of Transfer". The first phenomenon is because "transfer" is the independent root concept from "process", and "effect" is similar to the independent root concept that is separated from "effect". In this way, the "process" biological child has the characteristics of "transfer" ("legal", without having to set up in "transfer". This is better than the United States from the country of independence from the United Kingdom. The culture of the United Kingdom has a lot of commonality, and the research in many cultural fields can take advantage of this part (language philosophy). The second phenomenon comes with different basic types of "transfer" with different sentence clauses, and the different basic types of "process" do not have this feature. Such explanations are of course just "Taoustic" and not "Taoism,", if it is, because "transfer" is complicated by the time than "process", this one; "transfer" "Relationship" strong correlation, and "process" is associated with "relationship" weakly, this is two. The high-level design of each root concept has its own problems, here is not explained. Finally, it is necessary to emphasize that the high-level design of the complex primitive concept of human activities is summarized and performed, and this sub-space is the specificization of context. In the past, research on context mainly adopted "fox" mode. HNC changed with "hedgehog" mode, trying to give a computer to grasp the form of language mode. Of course, the composite primitive concept is just a symbolic foundation, the formation of the formation model is improved, and it has the same support for sentence groups, paragraphs, and chapters. There is also a deep cooperation problem of "fox" and "hedgehog". Maybe it can be said that the "marriage" of "Fox" and "Hedgehog" is the time when the computer can automatically generate contexts. Can Chinese linguistics walk in the world in this key area? It should be said that it is very hope!
(Fail)
