CN1838159B - Cognition logic machine and its information processing method - Google Patents
Cognition logic machine and its information processing method Download PDFInfo
- Publication number
- CN1838159B CN1838159B CN2006100548172A CN200610054817A CN1838159B CN 1838159 B CN1838159 B CN 1838159B CN 2006100548172 A CN2006100548172 A CN 2006100548172A CN 200610054817 A CN200610054817 A CN 200610054817A CN 1838159 B CN1838159 B CN 1838159B
- Authority
- CN
- China
- Prior art keywords
- information
- logic
- sign indicating
- indicating number
- consciousness
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 230000019771 cognition Effects 0.000 title claims abstract description 86
- 230000010365 information processing Effects 0.000 title claims abstract description 33
- 238000003672 processing method Methods 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims abstract description 200
- 230000008447 perception Effects 0.000 claims abstract description 189
- 230000015654 memory Effects 0.000 claims abstract description 138
- 230000008569 process Effects 0.000 claims abstract description 128
- 238000003860 storage Methods 0.000 claims abstract description 86
- 210000004556 brain Anatomy 0.000 claims abstract description 69
- 238000012545 processing Methods 0.000 claims description 242
- 230000001953 sensory effect Effects 0.000 claims description 158
- 238000013507 mapping Methods 0.000 claims description 73
- 230000035807 sensation Effects 0.000 claims description 65
- 238000004088 simulation Methods 0.000 claims description 22
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 238000012217 deletion Methods 0.000 claims description 10
- 230000037430 deletion Effects 0.000 claims description 10
- 230000001939 inductive effect Effects 0.000 claims description 9
- 230000009467 reduction Effects 0.000 claims description 9
- 230000003750 conditioning effect Effects 0.000 claims description 7
- 230000008676 import Effects 0.000 claims description 6
- 238000004886 process control Methods 0.000 claims description 6
- 208000037656 Respiratory Sounds Diseases 0.000 claims description 2
- 238000000151 deposition Methods 0.000 claims description 2
- 206010037833 rales Diseases 0.000 claims description 2
- 238000004148 unit process Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 9
- 230000006698 induction Effects 0.000 abstract 1
- 230000000007 visual effect Effects 0.000 description 45
- 230000006870 function Effects 0.000 description 22
- 210000002569 neuron Anatomy 0.000 description 19
- 230000001149 cognitive effect Effects 0.000 description 17
- 230000001537 neural effect Effects 0.000 description 17
- 210000001061 forehead Anatomy 0.000 description 14
- 230000002490 cerebral effect Effects 0.000 description 12
- 238000003058 natural language processing Methods 0.000 description 9
- 230000005540 biological transmission Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 239000000463 material Substances 0.000 description 7
- 229910052751 metal Inorganic materials 0.000 description 7
- 239000002184 metal Substances 0.000 description 7
- 210000000869 occipital lobe Anatomy 0.000 description 7
- 210000003478 temporal lobe Anatomy 0.000 description 7
- 238000013519 translation Methods 0.000 description 7
- 230000035479 physiological effects, processes and functions Effects 0.000 description 6
- 230000002360 prefrontal effect Effects 0.000 description 6
- 241000287828 Gallus gallus Species 0.000 description 5
- 241001465754 Metazoa Species 0.000 description 5
- 238000011161 development Methods 0.000 description 5
- 230000018109 developmental process Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000008520 organization Effects 0.000 description 5
- 241000272525 Anas platyrhynchos Species 0.000 description 4
- 240000007711 Peperomia pellucida Species 0.000 description 4
- 235000012364 Peperomia pellucida Nutrition 0.000 description 4
- 230000006399 behavior Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 4
- 210000004027 cell Anatomy 0.000 description 4
- 230000005611 electricity Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000010354 integration Effects 0.000 description 4
- 230000005055 memory storage Effects 0.000 description 4
- 206010010356 Congenital anomaly Diseases 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 3
- 238000004220 aggregation Methods 0.000 description 3
- 239000004411 aluminium Substances 0.000 description 3
- 229910052782 aluminium Inorganic materials 0.000 description 3
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 3
- 239000000835 fiber Substances 0.000 description 3
- 230000007787 long-term memory Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 210000003205 muscle Anatomy 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 241000282461 Canis lupus Species 0.000 description 2
- 244000025254 Cannabis sativa Species 0.000 description 2
- 241000283153 Cetacea Species 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 2
- 241000124008 Mammalia Species 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000003920 cognitive function Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 210000001652 frontal lobe Anatomy 0.000 description 2
- 239000004615 ingredient Substances 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 210000002442 prefrontal cortex Anatomy 0.000 description 2
- ATJFFYVFTNAWJD-UHFFFAOYSA-N Tin Chemical compound [Sn] ATJFFYVFTNAWJD-UHFFFAOYSA-N 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- WQAQPCDUOCURKW-UHFFFAOYSA-N butanethiol Chemical compound CCCCS WQAQPCDUOCURKW-UHFFFAOYSA-N 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000004720 cerebrum Anatomy 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000010924 continuous production Methods 0.000 description 1
- 210000003792 cranial nerve Anatomy 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 230000007334 memory performance Effects 0.000 description 1
- 210000000337 motor cortex Anatomy 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 230000036403 neuro physiology Effects 0.000 description 1
- 230000001936 parietal effect Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000008288 physiological mechanism Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000002910 structure generation Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Abstract
This invention relates to a cognition logic machine and its information processing method. Wherein, on deterministic Turing machine, it simulates hierarchic dual- memory structure of human brain and then generates logic cognition to relevant store and access natural language. The machine comprises: an input device, a logic cognitive process device, a knowledge-memory device, and an output device.Besides, the process of the machine comprises the perception, induction, deduction, and inversion, which realizes learning/forgetting, understand/ memory, recognition/image and reasoning in machine. This invention overcomes the problem of exponential growth for the command access algorithm time, and provides storage and inquest method for natural language process.
Description
Technical field
The present invention relates to a kind of memory structures and the cognition logic machine of epistemic logic information processing and method of its process information of simulating human brain, particularly relate to the method that this cognition logic machine is handled natural language.
Technical background
Existing information processing is mainly realized by computer system.Since first robot calculator is born, the development that machine information work of treatment acquisition is advanced by leaps and bounds.Itself has experienced electron tube computing machine, transistor, four epoch such as integrated circuit and large scale integrated circuit.On human History of Science ﹠ Technology, can mention in the same breath with the development of robot calculator without any subject.
" law of thinking " and the mathematical logic theory of its foundation of mathematician boolean in 1854 issue have been established the logical foundations of computer nowadays.Nineteen forty-six American Hungary mathematician Feng. Nuo Yiman has proposed the basic imagination about computing machine composition and working method.The computing machine that he proposed comprises arithmetical unit, storer, controller, the input and output device five big basic elements of character.Till now, although great variation has taken place in the computing mechanism technology for making, with regard to its architecture, still based on his design philosophy.
The working method of Feng. Nuo Yiman machine is: (1) controller is carried out according to the order of sequence according to the instruction ground of instruction of program that leaves in the storer.(2) every instruction comprises some operations, the substance of these operations be with data from a register after treatment to another register transfer.Each operation all is to finish under the order of controller.Because the theoretical foundation of Feng. Nuo Yiman machine is from the Turing machine model, this class computing machine is referred to as deterministic Turing machine.
Because the theoretical foundation of Feng. Nuo Yimanshi computing machine is mathematical logic, Computer Processing number reason problem now is incomparably successful.Yet unity and coherence in writing information such as processing natural language are but very difficult.Artificial intelligence approach at present commonly used, regular algorithm and statistic algorithm, all Protean natural language of processing form successfully.Present mathematical theory and computational practice show that the needed time of natural language processing is exponential increase with the complexity of input information.
People generally believe that the fifth-generation computer should be the machine of simulation human brain.Artificial intelligence theory's purpose allows computing machine work as the people exactly.The realization of artificial intelligence needs cranial nerve to learn, psychology, and logistics, linguistics and computer science fellowship provide unified and blanket model.
Patent " Memory system for a multi-chip digital computer " (3,821,751) with Hoff is example, and present computing machine belongs to " stored programme ", linear data disposal system.The working method of this computing system is the instruction and data that order is carried out string type.Its advantage is to handle numerical evaluation.Its inferior position is the relation storage of language message and handles, natural language statement especially, semanteme, sentence structure, understanding and the integration processing of asking for.
The research of computing machine is thought, has a kind of non-deterministic Turing machine (NondeterministicTuring Machine) in theory, and its principle of work is to use bilingual speech L1 and L2.If between the limited character string of two language, there is function f: ∑ * 1 → ∑ * 2, and make
The patent of Han (6,611,841) method is a kind of so bilingual speech, but the reduction machine technology.And can be engaged in the statement of natural language, and semanteme, sentence structure is understood and the integration processing of reasoning in polynomial time.But the method for natural language processing does not more specifically also have detailed description.
Summary of the invention
In order to overcome active computer (being also referred to as deterministic Turing machine) there is the difficulty that the exponential time increases in problem solvings such as natural language, the present invention proposes the machine that a kind of natural language epistemic logic of simulating human brain handles and the method for its process information.This machine comprises the epistemic logic of a cover anthropomorphic dummy brain.This epistemic logic is different from present mathematical logic, and it can be learnt by implementation language, understand, and cognitive functions such as reasoning and memory, and realize the relation storage of natural language and ask for processing.
A first aspect of the present invention proposes a kind of cognition logic machine of simulating the two memory structures of grade of human brain.This cognition logic machine comprises input media, bus, output unit, parts such as knowledge memory device and epistemic logic treating apparatus.Described input media, bus and output unit all can use input media, bus and the output unit of active computer.
Described knowledge memory device is a device that the relation memory of the two memories of grade is stored, and this pair memory comprises sensory storage and perceptual memory.The memory of this relation is meant the storage with relation information of feeling-knowing structure, and this relation storage concerns that by one array or one concern polynary group and realizes, for example polynary group of this relation can be one and concerns tlv triple (Tuple＜P/C, P, C 〉).This concerns that tlv triple comprises perception location sign indicating number (P/C), sensation location sign indicating number (P) and consciousness location sign indicating number (C).The address that this sensation location sign indicating number is used to store sensory information (hereinafter referred to as sensory information).This consciousness location sign indicating number is used to store the address of consciousness information (hereinafter referred to as consciousness information).This perception location sign indicating number can be used to store the address of perception information simultaneously, so the address that this perception location sign indicating number is a sensory information also is the address of consciousness information.Between this perception location sign indicating number and this sensation location sign indicating number or this perception location sign indicating number and this consciousness location sign indicating number all are mapping relations of one-to-many, by the sensory information of this perception location sign indicating number storage or consciousness information respectively and accordingly by the sensory information of sensation location sign indicating number storage or be the mapping relations of one-to-many between the consciousness information by the storage of consciousness location sign indicating number accordingly.Based on the interchangeability of perception location sign indicating number, promptly this perception location sign indicating number is sensory information the address also is the address of consciousness information, the mapping relations of the conversion character formation multi-to-multi by perception location sign indicating number between sensation location sign indicating number and the consciousness location sign indicating number.This perception location sign indicating number can generate according to physical message in the machine by a perception compilation unit.Physical message and extraneous input information are one to one in the described machine, and physical message also is one to one in this perception location sign indicating number and the machine.Utilize the above-mentioned tlv triple architectural feature that concerns, described cognition logic machine can be carried out the storage of sensory information and consciousness information mapping relation and ask for.
Cognition logic machine of the present invention also comprises an epistemic logic treating apparatus, and this epistemic logic treating apparatus is used for to external world input information and carries out logical process and convert thereof into having the relation information of feeling-know structure and storing the knowledge memory device into or output to output unit.The epistemic logic that this epistemic logic treating apparatus is utilized is different with the mathematical logic of traditional Turing machine model, this epistemic logic does not rely on state (State) to be programmed into the line function recursive operation, but the foundation and the inquiry that between sensory information and consciousness information two symmetric informations, concern.Wherein said epistemic logic can comprise the perception logic, inductive logic, deductive logic and the anti-logic of unraveling silk.Described perception logic is that input information is set up perception location sign indicating number to external world.Described inductive logic is the foundation that concerns between sensory information and the consciousness information.Described deductive logic is to ask for consciousness information from sensory information.The described anti-logic of unraveling silk is to ask for sensory information from consciousness information.Described epistemic logic treating apparatus can be carried out perception respectively by above-mentioned corresponding logic and handle, and concludes and handles, and deduces to handle and instead unravel silk processing.
Described epistemic logic treating apparatus further comprises attention mnemon, epistemic logic processing unit and grammar processing unit.The storage organization of wherein said attention mnemon and knowledge memory device is identical.Different is that attention mnemon is not done permanent storage, just is used for the various information of temporary processing procedure, and shutting down the temporary information in back will disappear automatically.
The epistemic logic processing unit comprises learns/forgets device, understanding/memory device and identification/imagination device.Learn/forget device and be used for the mapping relations between sensory information and the consciousness information are carried out stores processor and deletion processing, thereby realize the study of simulation human brain and the processing procedure of forgeing; Understanding/memory device be used for to the mapping relations between sensory information and the consciousness information deduce/instead unravel silk processing, thereby realize the understanding of simulation human brain and the processing procedure of memory; Identification/imagination device is used for looking, and listens, and the mapping relations between the sensory information of multiple extraneous shape information such as body and the consciousness information deduce/instead unravel silk processing, thus realize the identification of simulation human brain and the processing procedure of the imagination.
The epistemic logic processing unit can be many yards parallel comparators of a multidigit address code.Described epistemic logic processing unit carries out the set operation processing and feels-know conversion process the sensation location sign indicating number or the consciousness location sign indicating number of the perception location sign indicating number mapping of input under the control of grammar processing unit.It is to compare between two or more sets address codes that this set operation is handled, and the value that is equal to is as common factor, and incoordinate value is as difference set.The result of described process of aggregation is imported into the input information of noting can be used as in the mnemon next step multi-level set operation.Grammar processing unit the infologic processing that provides the control of general syntax (Universal Grammar) rule to instruct the epistemic logic processing unit is provided and leaves intermediate result in attention unit preparation further continuous information handling, and the result who handles exports the most at last.The general syntax rule control of this grammar processing unit comprises the vocabulary processing controls, phrase processing controls, meaning of a word processing controls, sentence structure processing controls, semantic processes control and knowledge processing control etc.Described grammar processing unit judges whether to carry out language message study storage or language and asks for processing according to the language message of input.If carry out the language learning storing process, described grammar processing unit can further judge whether to carry out study of words again according to the information of this input, phrase study, sentence structure study and lexical study etc.At described vocabulary, phrase, after the study of the meaning of a word and syntactic knowledge, described grammar processing unit is according to vocabulary, phrase, the meaning of a word and syntactic knowledge can be carried out semanteme generation.Above-mentioned various grammer controls are to handle subelement by the vocabulary of described grammar processing unit, and subelement handled in phrase, and subelement handled in the meaning of a word, and subelement, semantic processes subelement and knowledge processing subelement handled in sentence structure.Above-mentioned each processing subelement is realized.
The epistemic logic treating apparatus also comprises an ASCII edit cell, and the information translation that described ASCII edit cell is used for external world's input is a physical message in the machine, perhaps physical message in the machine is converted to extraneous output information.
The epistemic logic treating apparatus also comprises a perception compilation unit, and it is used for converting physical message in the machine to perception location sign indicating number, perhaps converts perception location sign indicating number in machine physical message.Physical message is one to one in this perception location sign indicating number and the machine.
The method that a second aspect of the present invention proposition utilizes cognition logic machine to carry out information processing, particularly this cognition logic machine carries out the method for natural language processing.Described method is intended to solve a series of problems that present natural language processing algorithm can't overcome, and comprises semantic singular problem, and the algorithm time is with the complexity index growing concern.
The present invention is based on the memory structures and the epistemic logic principle of human brain, and a kind of cognition logic machine of proposition concerns storage and the method for asking for to natural language information to natural language processing the time.
The method of this natural language processing is to utilize the general syntax rule control of natural language, carries out the epistemic logic processing to having the language message that concerns the triple store form.The operation that this epistemic logic is handled is carried out by the epistemic logic processing unit, and described general syntax rale store is in grammar processing unit.
At first the language message that the external world is imported converts to and has the linguistic relation information of feeling-know structure, and this translation process can utilize following perception to handle and realize.This linguistic relation information can be that the above-mentioned triple store form that concerns is stored the various relations between language (comprising natural language and the artificial language) information.Then, described epistemic logic signal conditioning package concerns that to described sensory information in the triple store form and consciousness information are carried out perception, conclusion, deduction, thereby anti-logical process such as unravel silk realizes learning/forgeing, understanding/memory, the function of the identification/imagination and reasoning.
Described perception handle be will input external information deposit the knowledge memory device in the form of perception location sign indicating number.
It is to set up mapping relations between sensory information and consciousness information that described conclusion is handled.
It is the process of asking for consciousness information mapping relation from sensory information that described deduction is handled.Concrete disposal route is in the mapping of the consciousness location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence.
Describedly instead unravel silk that to handle be process from consciousness requestor message sensory information mapping relations.Concrete disposal route is in the mapping of the sensation location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence.
It is to ask for consciousness information from sensory information that described reasoning is handled, and converts sensory information then to, carries out the process of asking for of consciousness information again.
Above-mentioned logical process process can adopt also, hands over, and is poor, changes, and four fundamental operations are operated and realized.
The practice significance that described perception is handled is with information conversion of equal value between state of feeling and consciousness state, is P=C with the perception formulate.For example P (ANIMAL)=C (ANIMAL) expression " ANIMAL " be sensory information also be consciousness information, can do conversion of equal value between them.From the Realization by Machine angle, perception be with same information stores in two addresses one to one, perhaps two of map addresses information one to one.Perception is basic cognitive function.
The practice significance that described conclusion is handled is the relation of setting up between sensory information and the consciousness information.With the conclusion formulate is P → C.For example P (DOG) → C (ANIMAL) represents the relation of " DOG " and " ANIMAL ".
The practice significance that described deduction is handled is to occur simultaneously from sensory information to ask for consciousness information, is P＞C with the deduction formulate.For example P (DOG, CAT, MOUSE)＞C (ANIMAL) expression occurs simultaneously from sensory information " DOG, CAT, MOUSE " and ask for out consciousness information " ANIMAL ".From realizing angle, the processing of deducing is many a reduction (≤m, many-onereducibility) computings from perception location sign indicating number to consciousness location sign indicating number.
The described counter practice significance of unraveling silk processing is to occur simultaneously from consciousness information to ask for sensory information, is P＜C with the anti-formulate of unraveling silk.For example P (CAT)＜C (PETANIMAL, CLIMB TREE) expression is asked for out sensory information " CAT " from consciousness information " PETANIMAL, CLIMB TREE " common factor.From realizing angle, instead unravel silk that to handle be to many reduction operations of feeling the location sign indicating number from perception location sign indicating number.
Described study/forget, understanding/memory, the identification/imagination and reasoning are described perception, conclude, and deduce and the anti-outside cognitive process of unraveling silk, this outside cognitive process is the external manifestation of handling with respect to above-mentioned various epistemic logics.
Described learning process is perception and inductive logic information process, so learning process also can be used perception and inductive logic formulate.For example, study " DOG " is that " ANIMAL " this process can be with perception formula and conclusion formulate
Perception formula: P (DOG)=C (DOG),
Perception formula: P (ANIMAL)=C (ANIMAL),
Conclude formula: P (DOG) → C (ANIMAL),
And utilize described formula to realize learning process by cognition logic machine.
Described understanding and identification are the deductive logic processing procedures, and therefore understanding and identifying also can be used the deductive logic formulate.Here understand the deductive procedure that is meant Word message, identification is meant the deductive procedure of acoustic information and image information.For example, " TREE ", " BUSH " and " GRASS " can be understood as " PLANT ", with the deduction formulate is
P(TREE，BUSH，GRASS)＞C(PLANT)。
The described memory and the imagination are the anti-logical process processes of unraveling silk, and therefore memory and imagination process also can be represented with the anti-logical formula of unraveling silk.Here recall the translation process that is meant Word message, the imagination is meant the anti-process of unraveling silk of acoustic information and graphical information.For example, can remember " WHALE ", be with the anti-formulate of unraveling silk according to " MAMMAL " and " LIVING IN OCEAN "
P(WHALE)＜C(MAMMAL，LIVING?IN?OCEAN)。
Described reasoning process is the process that continuous deduction is handled, so reasoning process can be with the deduction formulate
P (1)＞C (2)=P (2)＞C (3) is reduced to P (1)＞C (2)＞C (3).
For example P (POODLE)＞C (DOG)＞C (ANIMAL) expression " POODLE " is " DOG ", and " DOG " is " ANIMAL ".
When natural language information is handled, above-mentioned various logic is handled by the realization of epistemic logic treating apparatus, and the epistemic logic processing unit in this epistemic logic treating apparatus is realized above-mentioned various epistemic logic processing under general syntax (Universal Grammar) the rule control that grammar processing unit provides is instructed.Described grammar processing unit judges whether to carry out language message study storage according to language message input or language is asked for.If carry out the language learning process, described grammar processing unit judges whether that according to information needs carry out study of words again, phrase study, sentence structure study and lexical study etc.At described vocabulary, phrase, after the study of the meaning of a word and syntactic knowledge, described grammar processing unit is according to vocabulary, phrase, the meaning of a word and syntactic knowledge are carried out semanteme generation.
Described basic semantic generative process mainly is divided into three levels.First level is that level asked in part of speech, grammatical category information corresponding to each independent word of forming this nature statement in the natural statement that will import in this level carries out the processing of part of speech understanding/memory by the understanding in the epistemic logic processing unit/memory device, asks for out from the knowledge memory device and the information of asking for is deposited in the attention mnemon.Second level is that level asked in sentence pattern; In this level, to further carry out sentence/memory processing to depositing the part of speech of noting in the mnemon in the ground floor in by the understanding in the epistemic logic processing unit/memory device, from the knowledge memory device, ask for out corresponding sentence pattern, and deposit in the attention mnemon.
Tri-layer is the semantic level that generates; In this level, in above-mentioned two levels, deposited the natural statement of above-mentioned input in note in the mnemon sentence pattern and part of speech respectively, carry out sentence comprehension/memory by the understanding in the epistemic logic processing unit/memory device and handle the semantic structure statement that generation has syntactic structure, and it is deposited in the knowledge memory device.
Described grammar processing unit provides the control of general syntax (Universal Grammar) rule, and the vocabulary of natural language, phrase, the meaning of a word, statement and syntactic structure convert unified machine language rule control to.The control of described general syntax rule comprises the vocabulary processing controls, phrase processing controls, meaning of a word processing controls, sentence structure processing controls, semantic processes control and knowledge processing programmed control.Described grammar processing unit is the method that adopts level of information combination and resolution process control when carrying out above-mentioned control, and this method comprises disposal routes such as vocabulary, phrase, the meaning of a word, sentence structure and knowledge.Wherein said level of information combination is meant that elementary level information processing is delivered to the method for senior level information processing layer by layer.The method of described combination and resolution process control is that the relational query of making up and resolution process converts between the unified machine address sign indicating number is handled, being about to all input informations converts to by elementary to senior machine address sign indicating number by described epistemic logic, wherein the primary address sign indicating number is the unit that forms senior address code, and the relation between primary address sign indicating number and the senior address code is set up and inquiry is handled realization by described epistemic logic.
It is to adopt the level damascene structures that described sentence structure is handled, and described level damascene structures provides the storage mode of sentence structure and asks for mode.Described vocabulary is handled, and phrase is handled, meaning of a word processing, and sentence structure is handled, and whether semantic processes and knowledge processing are to learn/forget to handle or the processing of understanding/memory according to the content decision of language input.If learn/forget processing, this grammar processing unit is by learning/forget processing procedure and successively carry out study of words or forget, phrase study or forget, lexical study or forget, sentence structure study or forget, semantic study or forget with knowledge learning or forget.If understanding/memory is handled, this grammar processing unit is successively carried out vocabulary understanding by understanding/memory processing procedure or is recalled, and phrase is understood or recalled, and the meaning of a word is understood or memory, and sentence structure is understood or recalled semantic understanding or memory and knowledge understanding or memory.Described grammar processing unit itself is not carried out direct computing, and it instructs the epistemic logic processing unit to carry out logical process.
If described grammar processing unit judges that language message is input as language and asks for process, described grammar processing unit also input language is carried out the semanteme generation and sentence comprehension/memory is handled, and converts the result to external information output by described perception compilation unit and ASCII compilation unit.
Description of drawings
Figure 1 shows that human brain basic structure synoptic diagram;
Figure 2 shows that human brain information processing synoptic diagram;
Figure 3 shows that the message level illustraton of model;
Figure 4 shows that the information translation illustraton of model;
Figure 5 shows that cognition logic machine illustraton of model of the present invention;
Figure 6 shows that epistemic logic structural drawing of the present invention;
Figure 7 shows that the basic structure synoptic diagram of cognition logic machine of the present invention;
Figure 8 shows that the structural drawing of epistemic logic treating apparatus of the present invention;
Figure 9 shows that the study of cognition logic machine of the present invention/forget work principle figure;
Figure 10 shows that cognition logic machine of the present invention is at understanding/memory work principle figure;
Figure 11 a is depicted as the instance graph that cognition logic machine of the present invention carries out study of words;
Figure 11 b is depicted as the instance graph that cognition logic machine of the present invention carries out statement study;
Figure 11 c is depicted as the instance graph that cognition logic machine of the present invention carries out semantic study;
Figure 12 shows that the instance graph that cognition logic machine of the present invention is understood/recalled;
Figure 13 shows that cognition logic machine of the present invention carries out the semantic instance graph that generates;
Figure 14 a is depicted as the instance graph that cognition logic machine of the present invention carries out reasoning;
Figure 14 b is depicted as the instance graph of the reasoning principle of cognition logic machine of the present invention;
Figure 15 shows that many sensings epistemic logic structural drawing;
Figure 16 a is depicted as the structural drawing of the grammar processing unit logic of cognition logic machine of the present invention;
Figure 16 b is depicted as the instance graph of the grammer processing procedure of cognition logic machine of the present invention.
Embodiment
In order more clearly to introduce content of the present invention, now in conjunction with the accompanying drawings the present invention is elaborated, before introducing the present invention,, now the physiological structure and the congenital physiology logic of human brain are done following introduction in order to make content of the present invention easier to understand.
The two memory structures introductions of the grade of human brain
The structure of cognition logic machine and logical process rule are the two memory structures of grade of simulation human brain and the epistemic logic principle of work and power that this structure forms.
Figure 1 shows that the basic structure signal of human brain.Human brain is divided into the two cerebral hemispheres, and every hemisphere comprises four cerebral lobes.They are occipital lobe, temporal lobe, top and frontal lobe.The occipital ctx processing visual information, cortex of temporal lobe is handled auditory information, and parietal cortex is handled somatesthesia and cinesthesia information.We claim that this three leaves cortex is a sensory cortex, because the information that their sensations are obtained from physical world.Frontal lobe comprises two parts, motion parts and forehead portion.Motion parts is carried out the output of motion.Forehead portion is to handle the maincenter of abstracted information.
Theoretical scheme of the present invention is come according to people's brain neurophysiology theoretical developments.The cortex that each cerebral lobe of human brain is formed by neuron covers.The cortex of each cerebral lobe comprises three (elementary, intermediate and senior) cortex zones at least.Information passes to the middle rank district after preliminary area is processed, enter senior district again.After information is processed by these zones, become comprehensive complicated.For example, when visual information arrived elementary cortex by our brain, movable neuron can let us be felt and is seen simple bright spot in this zone.When visual information arrives senior cortex, perhaps can experience complicated sensed image by let us at the neuron of this regional activity, such as tree, animal, or the like.Similarly, when auditory information arrived elementary cortex, it can activate our the basic sense of hearing, and let us feels to hear simple sound equipment.When it arrived senior cortex, it can make us feel to hear that height changes or have the sound of joint rate.For the shake-up visual information of health, when it arrives at elementary cortex, can accurately activate shake-up consciousness.When it arrived senior cortex, it can activate whole shake-up consciousness.
In addition, the information of handling through perception is incorporated into the highest structure from each cerebral lobe, prefrontal area.Between prefrontal area and the sensation cerebral lobe nerve fibre connection is arranged.These fibers can roughly be divided into three groups of functions: the information transmission between forehead consciousness and occipital lobe visual sense are felt; Forehead consciousness and temporal lobe are listened the information transmission between the sensation; Information transmission between forehead consciousness and the top somesthesia.These fiber-like of at full speed transmitting information are similar to the cable of bunchy, come and go transmission information between prefrontal lobe and three sensation cerebral lobes.
Information is incorporated into the forehead brain from the sensation cerebral lobe and forms consciousness information.Can remember the visual sense visual information at occipital lobe by fibrous bundle in the consciousness information that the forehead brain forms, remember at temporal lobe and listen sensory information, remember somesthesia information at top.
Figure 2 shows that the signal that information is handled in the different levels of each cerebral lobe of brain.Visual sense feels that tin sensation and somesthesia information at first arrive the elementary zone of cortex, the premium area that enters cortex then by the intermediate zone of cortex.Directly import the forehead brain into and generate consciousness information there by fibrous bundle in the information that premium area is processed.Forehead brain consciousness information can be delivered to each sensation cerebral lobe by fibrous bundle, remembers the sensory information that is stored in the there, also can directly instruct motor cortex to carry out movable information output.
The volume brain is graduate pair of memory structures.Except that prefrontal area, the volume brain also comprises other three moving regions.They are primary motor areas, intermediate motor area and senior motor area.Frons is the five-star structure of human brain on function.It is considered to control the zone of intellection.Primary motor area is arranged simple muscle activity.Coordinate muscle activity the middle rank motor area.Senior motor area is integrated colony's muscle activity.Compare with the sensation cerebral lobe, information is machined into level of complexity from simple horizontal; And at the volume brain, information is to be assigned to primary level from level of aggregation.The highest compages are prefrontal cortex.The output of the information domination motion of prefrontal cortex.
Information processing shows that elementary cortex finishes under a kind of like this order: different low layer structure neurons gathers identical information and goes to identical high-level structure neuron.The neuron of high-level structure obtains similar information from different low layer structure neurons conversely speaking.Thus, information adds by each layer and may be classified man-hour.That is to say that brain information is classified in level, can be referred to as hierarchical classification again.Hierarchical classification mechanism not only is confined to elementary cortex.It is a basic principle of whole brain process information.
Physiological work shows that brain information is processed in a grading layer aggregated(particle) structure.That is to say that the neuron that carries identical information is connected to specific high-level structure from the low layer structure.In other words, if information has general character, the low layer structure neuron that carries information can pass to these common informations the specific neuron in the high-level structure, and forms new information pattern.This is the basic logic of brain structure processing cognitive information.At physiology, this logic is called as integration.The function of its demonstration low layer structure is controlled by the function of high-level structure.In other words, high-level structure is that the low layer structure is controlled in development.
When said information was processed in brain, it meaned that information passes to another group neuron from one group of neuron.When two groups of neurons are shared information, the neural tuple of different low layers to the neural tuple transmission of sharing share letter or the neural tuple shared never with the neural tuple of low layer extract the information of sharing.
Figure 3 shows that message level model signal in the human brain, this synoptic diagram shows when neuron is shared identical information, the neural tuple of different low layers gathers shared information to the neural tuple of sharing, and perhaps the neural tuple of Gong Xianging provides shared information to the neural tuple of different low layers.
Prefrontal area is one and shares neural tuple that it receives visual sense from corresponding sensory memory feels, listens sensation and somesthesia information.It directly obtains the visual sense visual information by coming and going fibrous bundle from occipital lobe, obtains from temporal lobe and listens sensory information.Obtain somesthesia information from top.Fibrous bundle transmission information is twocouese.That is to say that prefrontal area is also to occipital lobe temporal lobe and top mail message.
Figure 4 shows that information translation model synoptic diagram.When the visual sense visual information arrived occipital lobe, it activated the upper strata layer by layer and shares neural tuple, arrives the sensory information processing layer always.As shown in Figure 4, for example, a red circle and a yellow circle can activate the 5th group of neuron that they are shared at the visual processes layer.The 5th group of neuron of visual processes layer represented toroidal, will activate the neuron of visual expression layer, and continues to activate the neural tuple of consciousness layer ELCRIC.This sensory information can the neural tuple of voice activated information " circle ".This word spell backwards of ELCRIC is the difference that is used for showing between consciousness layer and the visual performance layer.
If see a green square for the first time, it will activate the 2nd and the 6th neural tuple, because they have sharing characteristic.Successively, these neural tuples will activate the 26th group of neuron of visual expression layer.The 26th group of neuron will activate the NEERG and the neural tuple of ERAUQS of consciousness level.The neural tuple of NEERG and ERAUQS will activate the neural tuple of representing this voice messaging of green square " green square ".This process is called as cognition.
The physiological structure of human brain provides a cognitive storage organization model.Top, the forehead brain is integrated from occipital lobe, and the visual sense of temporal lobe and top is felt, listens sensation and somesthesia information.The information of forehead brain also may be sent back to these sensation cerebral lobes.Two memory structures of grade and round information transmission are the bases of cognitive process.This is also for the invention provides epistemic logic principle basis.
Human cognitive is an integration system, and it is the product of human brain.According to the structural model of human brain, can summarize the unified epistemic logic principle of a cover, be called the natural intelligence Systems Theory.But characteristics of this natural intelligence system are its simulations, are in particular in:
1. human brain memory is the two memory systems of a Sensory-Perceptual grade.Sensory storage obtains information from the external world, represents presentation; Perceptual memory obtains information from sensory system, represents abstract.
2. sensory system mainly contains 3 kinds of forms: visual sense is felt, listens sensation and somesthesia.They can parallel processing or conversion under the guidance of consciousness system.
3. people's epistemic logic is the congenital physiology logic that produces between sensation and the two memory systems of consciousness grade, comprises perception, concludes, and deduces and counter unraveling silk.
4. perception is the same conversion between the Sensory-Perceptual information.
5. concluding is the foundation of Sensory-Perceptual information relationship.
6. deduction is the processing procedure that concerns from sensory information to consciousness information.
7. counter unraveling silk is the processing procedure that concerns from consciousness information to sensory information.
8. all cognitive process all is perception, concludes, and deduces and anti-combination of unraveling silk.
9. knowledge is the set of relationship between sensation and the consciousness information.
10. language is the expression of knowledge.
11. also, hand over, poor and to change be four basic operations of cognition.
12. epistemic logic is an architecture logic, can utilize machine simulation.
Here, memory is the basis of human cognitive, and the mankind come reservation information by it.Although some memory phenomenons for example, gain public acceptance with long-term memory in short-term, yet remain unknown with the physiological mechanisms of long-term memory and location in short-term.Different with traditional viewpoint, the epistemic logic theory thinks only be phenomenon with long-term memory in short-term.More precisely, Ren Lei cognitive information is present in two of presentation (sensation) and abstract (consciousness) each other in the memory system of mirror image.The two memory systems of this grade are canned data but also store two relations between the memory not only.Sensory storage is received information also and is had one to one between the material information from the material world and concerns.Perceptual memory is not directly received information from the material world, and receives information from sensory storage, and the relation of one-to-many is arranged between perceptual memory and the sensory storage.Thinking is carried out between sensory storage and perceptual memory.What this is illustrated as, and the people still can thinking when having the external information input.Give an example, if there is the people to tell us he has just found " black hole " theory, what we can understand that he talks about is what, and does not need to accept this notion from the material world.
Sensation and consciousness form partable, cooperative a pair of memory.We have similar experience: we can be appreciated that what we have forgotten sometimes, but what do not know that we forget is what.This example is the phenomenon of sensory storage disappearance.Sometimes we can forget the thing that others wants us to do, and can criticize the thing that others does not tell that we will do.When others took out physical evidence, we just found it is that oneself has been forgotten.This is the perceptual memory deficient phenomena.Physical evidence and sensory storage can be given perceptual memory for change and confirm what we have forgotten really.
Sensory storage has comprised three parts: sensation and somesthesia memory are felt, listened to visual sense.Sensory information was both also spatially handled in the time.Visual sense visual information processing list reveals spatial property, listens the sensory information processing list to reveal linear feature.Each sensory storage can be as the carrier of language.That is to say, carry the literal and the sound equipment symbol of language meaning and can ask for back consciousness information.Somesthesia is played the part of identical language carrier role in braille.These three kinds of sensory storages are connected to perceptual memory abreast.
Meaning of language is stored in the perceptual memory.Perceptual memory can be defined as the consciousness to sensory information.The relation object of perceptual memory and sensory storage is similar to electricity and magnetic is set up alternately, and is indispensable.On the other hand, the sensory storage performance is more passive.The perceptual memory performance more initiatively.Consciousness information is created from sensory information.Innovation and creation and other novel thought are the results that the new knowledge visual information generates new sensory information.The reason that increases with index speed of information content why in the world that Here it is.Although the mankind can not simulate the consciousness phenomenon, can simulate cognitive process.
Two-way processing between sensation and the consciousness information has formed the basis of congenital physiology logic.For example, the visual sense visual information of one group of relevant chicken, the image of for example white hen and flower cock gets up with the consciousness informational linkage of relevant chicken.Although about the visual sense of chicken feels that image is different on size and color, they in brain, be connected to one abstract, notional classification.The consciousness information of relevant chicken can be connected to the specific sensation of listening again to be remembered, for example one group of English visual symbol " Chicken ", the perhaps visual symbol of other languages or sound.When child learned word, abstract consciousness information was connected on the specific sensory information.
Use the same method, child can learn " bird ", " animal " and other words.That is to say that a people's vocabulary knowledge is exactly the contact between consciousness information and the sensory information in this human brain.When hearing or seeing a special word, will wake its relevant consciousness information up.
Consciousness information can be by being learnt with sensory information opening relationships.The two memory structures of grade provide a logic mechanism to come relation between the process information to people.For example: duck is related to bird, and bird is related to animal.To bird, bird is to existing continuous logical relation between the animal duck.If it is animal that people know bird, duck is a bird, so, just can reach a conclusion, and duck is an animal.In the term of epistemic logic technology, knowledge can be defined as relation between sensory information and the consciousness information by physiology.Sensory storage is passive relatively, and perceptual memory is relatively initiatively.Consciousness information is associated with consciousness and wish.Its instructor's behavior.Sensory information is associated with experience and physical message.The cognitive information that people accept from the material world mainly contains visual sense and feels, listens sensation and somesthesia form.The information of feeling is stored in sensory storage.
Experience the storage of visual sense visual information and listen the difference between the sensory information storage not difficult.For example, if require someone to recall the gate that he once attended school the middle school of book, he is easy to give relevant space visual sense visual information for change, because he is storing this visual sense visual information in brain.Acoustic memory be one discrete, be linear storage.If require him to recall first note of " March of the Volunteers " second trifle, he needs sequentially from memory " " dumbly to go out the note of first trifle, finds first note of second trifle then.As can be seen, the listen sensation and the spatial visual sense of timeliness feel different, and it is linear that storage is.
As if for common people, the somesthesia memory does not have visual sense to feel on communication and listens the sensation memory meaningful like that.The somesthesia memory is just essential for the blind person.The somesthesia memory can substitute " normally " language in needs.Braille comes to this, and development comes.The sportsman also relies on the somesthesia memory and goes to finish their sports.
Perceptual memory is different with sensory storage.Perceptual memory has the consciousness of discovering information.This is the memory function of level of consciousness.The consciousness phenomenon has exceeded the scope that this paper can discuss.This piece article only supposes that the memory of such level exists, and with the scientific approach proof with reappear cognitive phenomenon.But the two theory of memory of grade are supplied with the method for inspection that people understand human cognitive.Difference perceptual memory and sensory storage are easy to.For example, the gate of president Lincoln in Illinois man if require us to recall, the people who has not been the there does not have concrete visual sense visual information, but can know that in perceptual memory president Lincoln may have a family in the Illinois, and individual door is arranged.
Give an example again, if " ICSC " is presented in face of us, we seen after its can perception it, even can also recall it.If we do not know its how it feels, we just can not get up it with relevant concept connection.If we are apprised of " ICSC " is " ICSC ", its network address is www.icsc.org, and we can connect consciousness information and " ICSC ", simultaneously, also can transmit these information and give other people.
Cognitive phenomenon shows that information is stored and handles on two levels.When our one of perception did not have the information of implication, it was stored in a pair of sensation and the perceptual memory.When we know the implication of sensory information, represent the consciousness information of implication and sensory information just to set up relation.Knowledge is the set of the relation between sensory information and the consciousness information.If the relation between " ICSC " and " ICSC " is set up in the brains at us, we can give this specific knowledge for change.Otherwise, " ICSC " for us without any meaning.
Knowledge is made of four important elements: symbol, implication, relation and time.Sensory storage is handled and stored symbols.Perceptual memory is realized on the basis of time, handles relation and meaning between the sensory information, and stores them.Cross over the basis that time relation has constituted knowledge between sensory information and the consciousness information.
When handling knowledge, consciousness information is presented with linear mode by sensory information.The order of sensory information stream is followed temporal mode.This pattern is exactly a grammer.Grammer is the fundamental characteristics of language.Use the people of identical language following with a kind of syntax rule.Different language can have different grammers.Children are by themselves the notion generative capacity rule that learns grammar.This born ability will come into question in ensuing part.Handling knowledge is the process of thinking.Knowledge can obtain by language, because language is comprising word and sentence, and word and sentence are comprising the grammer of semantic and time sequencing.
Language dark, shallow double-decker is accepted quite at large by linguistic circles., how these two structures work together, what physiological structure to be made of not known again.For the interpretative code phenomenon, the linguist recognizes double-deck necessity.Chomsky has been introduced the canonical form of deep structure and has been judged that different syntactic structures can comprise identical proposition and judge.For example, " John has write envelope letter to the Mary " and " envelope is write to Mary's letter and is write by John ", these two words are meanings, and different sentence structurees and different shallow structure are but arranged.
The theory that sentence is handled between sensation and two levels of consciousness makes language phenomenon be easy to be explained.Word and word order that the consciousness meaning of each sentence can be experienced determine.Read or listen then when sentence, its word order can be recognized.This order can help us that the meaning of each speech is combined in the specific structure to go.The meaning that different structures shape is different.In addition, have under the situation of a plurality of implications at a word, the selection specific meanings depends on the relation between the word.For example, the implication of Java in sentence " Aprogrammer is writing Java " and " A traveler is visiting Java " is respectively by word writing and visiting decision.
It is born understanding a string symbol and using the ability of a notion of a string symbolic representation.The rule of word order is called as grammer.Different language has different grammers.The knowledge of grammar is not born, and, the ability of being familiar with these sentence patterns is born.In other words, children can just use grammer before they formally learn these syntax rules.The linguist claims this connate ability to be unified grammer.Carry out extensive studies for unified grammer, but also do not had to form the theory that to verify at present.
Notion entered human brain sometimes before language, and then described in words it.Word computer and internet introduce after notion is developed.That is to say that consciousness information is independent of the material language.Notion and sensory information are processed in double-layer structure, and connect each other between them.
The ability that expresses thoughts of utilization language can be explained with sensation and the two memory structures of consciousness and the information transmission between them.Semanteme is the phenomenon that language has implication.When we do definition for word, can find own at other word of use.Why we must be fans with other notions of some concept explanations all the time.This process can be explained like this: when the symbol of language, as APPLE, when being learnt, it is associated with specific consciousness information, thereby this specific consciousness information becomes the meaning of APPLE.This phy symbol can be associated with its conceptual in brains.Conceptual also can be given its sensation contact of performance for change.The combination of symbol also can be found the combination of conceptual, and vice versa.
The symbol combination mode has determined the meaning of symbols, thus the constituent grammar phenomenon.Just as we have described, grammer is how to make up to present the knowledge of different implications about word.Just, the order of group of words is determining different consciousness information.For instance, " wolf chases after dog " has different implications with " dog chases after wolf ".The order energy quilt of word automatically consciousness arrives, and is used for determining the implication of sentence.
Therefore, according to the physiology principle, the two memory structures of the grade of human brain have formed structural flogic system.Sensory storage obtains information from the material world, and perceptual memory obtains information from sensory storage.Sensory storage can be thought the surface structure of language, and perceptual memory can be thought the deep structure of language.Knowledge is the multiple information relationship between these two memories.Information translation between these two memory structures forms a cover cognitive information logic principle.
Recently Nervous System Anatomy confirms that to the research of memory human brain is comprising a dual structure system really.Sensory storage and perceptual memory are two kinds of independently cerebral functions.Other neuroanatomical research points out that people's forehead brain is relevant with conceptual memory, and sensory cortex is carried out the sensory information storage.The fiber that connects forehead brain and sensory cortex forms thick nerve tract.Information comes and goes between forehead brain and sensory cortex by this high speed information cable and transmits.The two memory structures of this grade form the epistemic logic basis.
The present invention is based on the memory structures and the epistemic logic principle of the above-mentioned human brain of simulation, a kind of two memory structures of grade of simulating human brain that propose reach the machine and the method for the inherent epistemic logic of structure generation thus, and with this machine and method natural language are concerned storage and the processing of asking for.Figure 5 shows that the model of cognition logic machine of the present invention.Relation memory epistemic logic among the present invention comprises sensory information 101 and consciousness information 102 and handles the epistemic logic treating apparatus 103 that concerns between the described pair of recall info.Different with traditional Turing machine model, this not dependence of epistemic logic disposal system model state is programmed into the line function recursive operation, but the foundation and the inquiry that between the language of sensory information and two symmetries of consciousness information, concern.
Figure 6 shows that the structural model of epistemic logic of the present invention.Statement theoretically, described epistemic logic treating apparatus 103 comprises perception processing unit 104, deduction processing unit 105 and the anti-processing unit 106 of unraveling silk.Described epistemic logic treating apparatus 103 is used for to external world input information to carry out logical process and converts thereof into and have the relation information of feeling-know structure and store the two memory storage of grade into.The two memory storage modes of this grade are storages of a kind of relation.Specifically, this storage mode is the storage with sensory information and consciousness information mutual mapping relation.Described epistemic logic treating apparatus carries out logical process according to mapping relations between 100 pairs of sensory informations 101 of external information input and the consciousness information 102 and the result that will handle is input in the grade pair memory storage, perhaps outputs to output unit.
Figure 7 shows that the basic structure synoptic diagram of cognition logic machine of the present invention.This cognition logic machine comprises input media 111, bus 112, epistemic logic treating apparatus 103, knowledge memory device 115 and output unit 116.
Be illustrated in figure 8 as the structural drawing of the epistemic logic treating apparatus in this cognition logic machine shown in Figure 7.Because it is different with the data storage memory of traditional Turing machine model that the data storage of the cognition logic machine among the present invention is remembered, the storage mode among the present invention is the storage to sensory information and consciousness information mutual mapping relation.Wherein said epistemic logic treating apparatus carries out the processing of epistemic logic, comprises perception, concludes, and deduces and anti-processing such as unravels silk.The epistemic logic of epistemic logic treating apparatus is different with the mathematical logic of traditional Turing machine model, epistemic logic does not rely on state (State) to be programmed into the line function recursive operation, but the foundation and the inquiry that between the language of sensory information and two symmetries of consciousness information, concern.The perception logic is the correspondence one by one between sensory information and the consciousness information.Inductive logic is the foundation that concerns between sensory information and the consciousness information.Deductive logic is to ask for consciousness information from sensory information.The anti-logic of unraveling silk is from consciousness requestor message sensory information.
Epistemic logic signal conditioning package 103 of the present invention comprises epistemic logic processing unit 113, grammar processing unit 127 and attention mnemon 114.Wherein said epistemic logic processing unit 113 further comprises learns/forgets device 123, understanding/memory device 124 and identification/imagination device 125.Described study/forget device 123 is used for the mapping relations between described sensory information 101 and the consciousness information 102 are stored and delete handling and realize the study of simulation human brain and the processing procedure of forgeing; Described understanding/memory device 124 be used for to the mapping relations between described sensory information 101 and the consciousness information 102 deduce/counter unraveling silk handle and realize the understanding of simulation human brain and the processing procedure of memory; Described identification/imagination device 125 is used for looking described, listens, and the mapping relations between somesthesia information 101 and the consciousness information 102 deduce/and counter unraveling silk handle and realize the identification of simulation human brain and the processing procedure of the imagination.
Described grammar processing unit 127 the infologic processing that provides the control of general syntax (Universal Grammar) rule to instruct epistemic logic processing unit 113 is provided and leaves intermediate result in attention unit 114 carrying out further continuous information processing, the result who handles the most at last output.Described grammar processing unit comprises vocabulary processing subelement 173, and subelement 174 handled in phrase, and subelement 175 handled in the meaning of a word, and subelement 176, semantic processes subelement 177 and knowledge processing subelement 172 handled in sentence structure.Described grammar processing unit judges whether to carry out language message study storage according to language message input 170 or language is asked for.If carry out the language learning process, described grammar processing unit judges whether that according to information needs carry out study of words again, phrase study, sentence structure study and lexical study etc.At described vocabulary, phrase, after the study of the meaning of a word and syntactic knowledge, described grammar processing unit 127 is according to vocabulary, phrase, the meaning of a word and syntactic knowledge are carried out semanteme generation 143.
Described epistemic logic processing unit 113 can be many yards parallel comparators of a multidigit address code.The sensation location sign indicating number of the perception location sign indicating number mapping of 113 pairs of inputs of epistemic logic processing unit under the control of grammar processing unit 127 or consciousness location sign indicating number carry out set operation and sense-know conversion process.This set operation is handled and is compared between two or more sets address codes, and the value that is equal to is as common factor, and incoordinate value is as difference set.The result of process of aggregation is output to the input information of noting in the memory 114 as next step multi-level set operation.For example, the sensation location sign indicating number " 1011,1100 " of perception location sign indicating number " 1000 " mapping is with the sensation location sign indicating number " 1011; 1110 " of perception location sign indicating number " 1010 " mapping, common factor be 1011, be expressed as P (1011)＜C (1000,1010), wherein P (1011,1100) → and C (1000), P (1011,1110) → C (1010).The consciousness location sign indicating number " 1101; 1111 " of perception location sign indicating number " 1000 " mapping is 1111 with the common factor of the consciousness location sign indicating number " 1110; 1111 " of perception location sign indicating number " 1001 " mapping, be expressed as P (1000,1001)＞C (1111), wherein P (1000) → C (1101,1111), P (1001) → C (1110,1111).As seen, the computing of epistemic logic processing unit is a reduction operations here.
Described cognition logic machine also comprises an ASCII edit cell, ASCII edit cell 131 as shown in Figure 9, and the information translation that is used for external world's input is a physical message sign indicating number in the machine, perhaps physical message sign indicating number 130 in the machine is converted to extraneous output information.Physical message is one to one in this ASCII and the machine.
Described cognition logic machine also comprises a perception compilation unit 132, and it is used for converting physical message in the machine 130 to perception location sign indicating number 104, perhaps converts perception location sign indicating number in machine physical message 130.This perception location sign indicating number 104 is one to one with the interior physical message 130 of machine.
Described attention mnemon 114 is identical with the storage organization of knowledge memory device.Different is that attention mnemon 114 is not done permanent storage, shuts down back knowledge information and disappears automatically.
Another aspect of the present invention proposes the method for cognition logic machine process information of the present invention, is described in detail below in conjunction with the principle of work and the information processing method of accompanying drawing to epistemic logic device of the present invention.The disposal route of described cognition logic machine mainly comprises the steps: the external information input step, the treatment step of epistemic logic treating apparatus, the storing step of knowledge memory device and external information output step.
The external information input step:
The input media of cognition logic machine of the present invention is set input mode according to the kind and the demand of external information 100, and external information is imported cognition logic machine, and formation and this external information be the interior physical message 130 of this cognition logic machine one to one.
Wherein, kind according to the information of external world input can be divided into stratum level with corresponding physical information, for example: rudimentary physical message, elementary physical message, the middle rank physical message, advanced physical information or the like, wherein the pairing physical message 130 of rudimentary unit with external information 100 is the physical message of rudimentary unit, for example, the English text information of input, each letter can be used as the rudimentary unit of the external information of input, and the physical message that each alphabetical pairing physical message is rudimentary unit.Elementary physical message is the set by a plurality of rudimentary units physical message.For example, the pairing physical message of each word of input can be regarded as elementary physical message, and it is the set that is made of rudimentary unit physical message letter.The middle rank physical message is the set that is made of elementary physical message.For example, the set that pairing elementary physical message word constitutes can be regarded as in the sentence of input.Advanced physical information can be made of intermediate physical message.The rudimentary unit of wherein said external information can be by system's automatic setting, therefore with the corresponding rudimentary unit of external information physical message, and elementary physical message, intermediate physical message and advanced physical information etc. can be set according to specific circumstances.
Described external information 100 can convert perception information 104 to by physical message in the machine 130.For example, video information can convert the visual sense visual information to, and audio-frequency information can convert to listens sensory information, and information that temperature sensor sends and cinesthesia information can convert sense of touch visual information etc. to such as the movable information that knocks keyboard.The mode of this input can be set according to the kind of information and the requirement of cognition logic machine, for example, and keyboard entry method, phonetic entry mode, sensor input mode and pattern input mode etc.
Described physical message can be set according to demand.The information that for example adopts the mode of keyboard typing to import can directly adopt ASCII character.
As shown in Figure 9, by keyboard successively with external information (E) 100, word " DOG " for example, and word " ANIMAL " is input in the cognition logic machine, because this input is to adopt the keyboard input, so can utilize existing ASCII character that the input word is handled one to one, and physical message (PH) 130 in the formation machine.In above-mentioned input, alphabetical pairing physical message can be used as the physical message of rudimentary unit, and for example alphabetical " D ", physical message can be regarded the physical message of rudimentary unit as in " O ", " G " pairing machine.Physical message can be seen elementary physical message as in the corresponding machine of word " DOG " institute.Generally speaking, low-level physical message constitutes high level physical message.This point also is understood that according to the custom of input.Word formed in letter, letter becomes sentence with group of words, constituting and the corresponding elementary physical message of word of same and alphabetical corresponding rudimentary unit physical message is with letter and word corresponding rudimentary unit physical message and elementary physical message formation and corresponding middle rank of sentence and advanced physical information.For example, physical message (PH) 100010010011111000111 can be regarded as by physical message 1000100 in the machine of letter " D " in the machine of " DOG ", and the interior physical message 1000111 of the machine of physical message 1001111 and " G " constitutes in the machine of " O ".
The treatment step of epistemic logic treating apparatus:
As shown in Figure 8, epistemic logic treating apparatus 103 comprises grammar processing unit 127, notes mnemon 114 and epistemic logic processing unit 113.Described epistemic logic processing unit 113 is a common comparer on function, handles the friendship of physics location sign indicating number information, and, difference and change computing.Describedly know that Logical processing unit 113 comprises and learn/forget device 123, understanding/memory device 124 and identification/imagination device 125.The information that described epistemic logic processing unit 113 is imported is to external world carried out the epistemic logic processing, and this processing comprises learns/forget a pair of negative function that device 123 is carried out.Learning functionality is to set up the function of sensory storage 101 and perceptual memory 102 relations.Forgeing function is the function of deletion sensory storage 101 and perceptual memory 102 relations.Understanding/memory device 124 is carried out and is understood and recall and handle, and identification/imagination device is carried out the processing of the identification and the imagination.Described understanding and memory are handled and the processing of the identification/imagination also is a pair of reverse each other function, and the input information of the further multi-level processing of do in the attention mnemon 114 is carried out and intermediate result is temporarily deposited in to this result under the guidance of grammar processing unit 127.
To describe this epistemic logic treating apparatus below in detail and be and how realize above-mentioned study/forget, understanding/memory, the identification/imagination is handled.
Learn/forget processing:
As shown in Figure 9, epistemic logic treating apparatus 103 comprises grammar processing unit 127, learns/forget device 123, notes mnemon 114, perception compilation unit 132 and ASCII compilation unit 131.
This epistemic logic signal conditioning package 103 by the study handling procedure to external world input information set up and to have the relation information that concerns tlv triple 115 file layouts.Figure 11 a is depicted as the study of words example, and this concerns that tlv triple 115 file layouts comprise perception location sign indicating number (P/C) 104, sensation location sign indicating number (P) 101 and consciousness location sign indicating number (C) 102.This external information (E) the 100th is input in the system by input media 111.Shown in Figure 11 a, outside lexical information " DOG " and " ANIMAL " are input in the cognition logic machine of the present invention by keyboard, by shown in the corresponding machines of the above-mentioned letter compiling of 131 pairs of ASCII compilation units in physical messages (PH) 130.This external information (E) is man-to-man mapping relations with the mapping relations 131 of the interior physical message (PH) of machine.Here, physical message is 1000100 in " D " pairing machine, and physical message is 100010010011111000111 in " DOG " pairing machine.Physical message can be divided into alphabetical rank, vocabulary rank and statement rank etc. in the machine.Physical message 1000100 belongs to alphabetical rank in the machine.Physical message 100010010011111000111 belongs to the vocabulary rank in the machine.Described level other divide relevant with the type of the external information of importing.
Then, physical message (PH) 130 is carried out the perception processing in the machine of 132 pairs of inputs of perception compilation unit of described epistemic logic signal conditioning package 103, promptly carries out the coding and the decoding of perception location sign indicating number (P/C) 104 one to one.Described perception location sign indicating number (P/C) 104 is man-to-man mapping relations with the mapping relations 132 of the interior physical message (PH) 100 of machine.This perception location sign indicating number can be set at random according to the demand of system.When this perception is handled, judge at first whether physical message has created this perception location sign indicating number in this machine.If do not set up this perception location sign indicating number, this perception compilation unit 132 will compile a perception location sign indicating number to it.This perception location sign indicating number can be automatically along the natural number of compiling according to the input sequence of physical message sign indicating number.In Figure 10 a, the perception location sign indicating number of letter " D " is 4, and the perception location sign indicating number of " DOG " is 103.If set up this perception location sign indicating number, this perception compilation unit 132 will use this perception location sign indicating number.
Subsequently, the study of described epistemic logic signal conditioning package 103/forget device 123 between sensory information and consciousness information, to set up (or deletion) mapping relations.The mapping relations of being set up are embodied by the relation of the tuple between pairing perception location sign indicating number (P/C) 104 and the consciousness location sign indicating number (C) 102.Thereby the activity of simulation human brain realizes study or forgets.The study here is the process of setting up the mapping relations between sensory information and the consciousness information, and forgeing is the process of the mapping relations between deletion sensory information and the consciousness information.Shown in Figure 11 a, when studying English " DOG " this word, whether perception compilation unit 132 at first compiled perception location sign indicating number to physical message (PH) in the identical machine to physical message (PH) retrieval in the above-mentioned machine, so that problem one to one appears not in interior physical message of the machine that prevents and perception location sign indicating number.If do not have to find physical message in the pairing machine of above-mentioned word was compiled perception location sign indicating number, the perception compilation unit is 103 to its compiling perception location sign indicating number here.In its correspondence sensation location sign indicating number memory 101, add 4,15 and 7 sensation location sign indicating number relations simultaneously as perception location sign indicating number 103.Can be expressed as, and P (D, O, G) → C (DOG).Equally, the perception location sign indicating number 203 of word " ANMIAL " can with letter " A ", " N ", " I ", " M ", sensation location sign indicating number 1,14,9,13,1 and 12 opening relationships of " A " and " L ", can be expressed as P (A, N, I, M, A, L) → C (ANIMAL).115 pairs of above-mentioned perception location sign indicating numbers of knowledge memory device of the present invention 104 are stored with the mapping relations of sensation location sign indicating number 101.
Above example is the learning process of vocabulary knowledge, below respectively to knowledge concerning word sense, and statement knowledge, the learning process of syntactic knowledge and semantic knowledge describes.
Figure 11 b is depicted as the study example that cognition logic machine of the present invention carries out statement knowledge." POODLE IS A KIND OF DOG " is input in the system of the present invention when external statement information, ASCII compilation unit 131 has compiled physical message (PH) 130 in the machine (having omitted physical message in the concrete machine among the figure) to each external information, this external information 100 is man-to-man mapping relations with the mapping relations 131 of the interior physical message of machine, 132 pairs of perception compilation units above-mentioned each do not compile that physical message compiles corresponding perception location sign indicating number 104 in the machine of perception location sign indicating number, and the mapping relations 132 of physical message and perception location sign indicating number also are man-to-man mapping relations in the machine.Here the perception location sign indicating number of " POODLE " is 101, the perception location sign indicating number of " IS " is 102, the perception location sign indicating number of phrase " A KIND OF " is 222, the perception location sign indicating number of word " DOG " be 103 and the perception location sign indicating number of sentence " POODLE IS A KIND OF DOG " be 301.The perception location sign indicating number 104 of above external information is all stored at knowledge memory device 115 of the present invention with sensation location sign indicating number 101 or with the mapping relations of consciousness location sign indicating number (C) 102.
Generally we compile perception location sign indicating number according to the order of the input of external information, preferred a kind of compile mode is the type in conjunction with the input external information, be that the rank of physical message compiles corresponding perception in the machine, for example above-mentioned word " POOPLE ", what " IS " and " DOG " was corresponding all is physical message in the elementary machine, so its perception location sign indicating number can be compiled as same level other 101,102 and 103.Phrase " AKIND
OF " rank of physical message is a physical message in the intermediate machine in the corresponding machine; so its perception location sign indicating number is 201; and the rank of physical message is a physical message in the senior machine in the pairing machine of statement " POODLE IS A KIND OF DOG ", its perception location sign indicating number is 301.Said sensed location sign indicating number also can be divided and store when the knowledge memory device is stored according to above-mentioned rank.
The external information that the present invention learnt/forgot 123 pairs of all inputs of device is set up perception relation between the information by physical message in the machine.For example, sentence " POODLE IS AKIND OF DOG " the words is as consciousness information, and its sensation relation is to be made of corresponding word and expression, i.e. " POODLE ", " IS ", " AKIND OF " and " DOG ".That is to say that the word and expression of sentence and composition sentence constitutes the mapping relations of the one-to-many between consciousness and the sensory information.Above-mentioned conversely speaking, word and expression is the sensation ingredient of this sentence.So the word of above-mentioned composition sentence and phrase and this sentence are many-to-one Sensory-Perceptual mapping relations.Above-mentioned mapping relations are embodied in perception location sign indicating number 104 and sensation location sign indicating number 101, and the expression-form between perception location sign indicating number 104 and the consciousness location sign indicating number 102 is: the perception location sign indicating number of sentence " POODLE IS AKINDOF DOG " is 301.Word " POODLE ", " IS ", the perception location sign indicating number of " DOG " and phrase " A KINDOF DOG " is 101,102,103 and 222.Sentence " POODLE IS A KINDOF DOG " and word " POODLE ", " IS ", the relation of " DOG " and phrase " A KIND OF DOG " can be with concerning that triple table is shown P (101,102,222,103) → C (301).Convert to external information can be expressed as P (POODLE, IS, A KIND OF, DOG) → C (POODLE IS AKIND OF DOG).
Figure 11 c is depicted as the study example that cognition logic machine of the present invention carries out semantic knowledge.Simultaneously referring to Figure 13, the semantic example that generates, during while statement input cognition logic machine, grammar processing unit 127 is handled the part of speech and the sentence pattern of statement according to the knowledge of grammar, and generates the semanteme 143 with syntactic structure according to sentence pattern.The grammer processing procedure will be introduced below in more detail.Natural language does not mark syntactic structure on literal, because the processing of syntactic structure is carried out in the consciousness level.For example during the described cognition logic machine of statement " POODLE ISAKIND OF DOG " input, except statement itself and the vocabulary that constitutes statement deposit in the knowledge memory device 115 according to perception relation rule mentioned above, grammar processing unit 127 is asked for out the part of speech of each vocabulary of statement " POODLE IS AKIND OF DOG " by understanding/memory device 124 according to the knowledge of grammar.Be formulated as P (POODLE IS A KIND OF DOG)＞C (NOUN, BE, PHRASE, NOUN)=P (NOUN, BE, PHRASE, NOUN)＜C (WORD CLASS).The resulting result of understanding/memory process deposits in and notes memory 114.Understanding/memory process will be introduced hereinafter in more detail.
Described grammar processing unit 127 continues to utilize the part of speech knowledge that leaves in the attention memory 114 to ask for the sentence pattern of correspondence " NOUN, BE, PHRASE, NOUN " by understanding/memory 124 according to the knowledge of grammar.Be formulated as
P(NOUN，BE，PHRASE，NOUN)＞C((){}\/[])＝P((){}\/[])＜C(SYNTAX)。Wherein " () { }/[] " is used for representing subject, predicate, complement and object etc.The sentence pattern knowledge of asking for temporarily deposits in notes memory 114.
Under the control of described grammar processing unit 127, the statement " POODLE IS A KIND OFDOG " of input embed corresponding syntactic knowledge C in the described attention memory 114 (() { }/[]) generative semantics knowledge C ((POODLE) IS} A KIND OF/[DOG]), and deposit in the knowledge memory device 115.
With reference to Figure 11 c, described semantic generative process is formulated as: P ((POODLE IS A KIND OF DOG), (() { }/[])) → C ((POODLE) IS} AKIND OF/[DOG])
Described built-in function can be formulated as:
P((101，211，222，202)，((){}\/[]))→C(255，101，256，265，211，266，275，222，202，276)
Above-mentioned several situation is the vocabulary with natural language, the explanation that statement and semanteme etc. is done the learning processing of epistemic logic treating apparatus of the present invention for example.Certainly above-mentioned example only is used for the learning processing explanation rather than is used for limiting this learning processing.In addition, above-mentioned learning processing can be finished also in the inside of this cognition logic machine and can directly deposit in the knowledge memory memory storage of this system the outside of this system is set up described cognitive relation in advance to external information after.
The forgeing of epistemic logic treating apparatus of the present invention handled and to be and opposite a kind of processing is handled in study, its be used for deleting perception information and sensory information or and consciousness information between the process that concerns of cognition.Because this processing is readily appreciated that, only be to learn/forget device will be stored in relevant perception information and sensory information in the knowledge memory device as requested, so the perhaps relation of the cognition between perception information and consciousness information deletion is no longer detailed description.Here, when doing the deletion processing, can be as requested only with the sensory information of perception information and mapping, perhaps the mapping relations of perception information and consciousness information are done the deletion processing; Also can be only with perception information, sensory information and consciousness information itself are done deletion.This can select according to requirement in practical systems.
Understanding/memory is handled:
Figure 10 shows that the fundamental diagram that epistemic logic device execution understanding/memory of the present invention is handled.It is of the present invention that to understand processing procedure be the process of being asked for consciousness information by sensory information.Memory processing procedure of the present invention is the process by consciousness requestor message sensory information.Understanding of the present invention/memory process is asked for specific consciousness location sign indicating number according to given perception location sign indicating number respectively, is expressed as P＞C, or asks for specific sensation location sign indicating number according to given perception location sign indicating number, is expressed as P＜C.It is to carry out according to the syntax rule in the grammar processing unit 127 that relation information is asked for and set operation that understanding/memory is handled.External information has been set up sensory information and consciousness information relationship by above-mentioned learning process and this relation has been stored in the knowledge memory device.Understanding/memory processor 124 is asked for the row set computing of going forward side by side of corresponding sensation location sign indicating number or consciousness location sign indicating number according to the perception location sign indicating number of input information by the syntax rule in the grammar processing unit 127 from the knowledge memory device, and the information of asking for of transition state temporarily is kept at notes in the mnemon 114 as the input information of further asking for.
Figure 12 shows that the example that cognition logic machine of the present invention is understood/recalled, for example, the part of speech of read statement " POODLE IS A KIND OF DOG " can be asked for from the knowledge memory device by understanding/memory device 124.Referring to Figure 13, the part of speech of statement " POODLE IS AKIND OF DOG " is asked for by understanding/memory processing 141 and is obtained NOUN (noun), BE (verb), PHRASE (phrase), and NOUN (noun) simultaneously.Be formulated as P (POODLE IS A KIND OF DOG)＞C (NOUN, BE, PHRASE, NOUN)=P (NOUN, BE, PHRASE, NOUN)＜C (WORD CLASS).Its inner location sign indicating number computing is P (101,211,222,202)＞C (241,242,243,241)=P (241,242,243,241)＜C (223).
Same reason, the sentence pattern of statement " POODLE IS AKIND OF DOG " also can be handled by understanding/memory and ask for shown in 142 among Figure 13, obtain () { }/[].Be formulated as
P(NOUN，BE，PHRASE，NOUN)＞C(()，{}，\/，[])＝P(()，{}，\/，[])<C(SYNTAX)。
Reasoning is handled:
Referring to Figure 14 a, b, described inference function is continuous understanding process, is formulated as
P(1)＞C(2)＞C(3)。For example, statement knowledge " aluminium is metal " and " metal can conduct electricity " can be stored in the knowledge memory device of cognition logic machine of the present invention by learning functionality.Be formulated as
P (aluminium) → C (metal) P (01) → C (02)
P (metal) → C (metal can conduct electricity) P (02) → C (31)
P (metal)=C (metal) P (02)=C (02)
Thereby can reasoning draw
P (aluminium)＞C (can conduct electricity) P (01)＞C (31)
Identification/the imagination is handled:
Referring to Figure 15, it is that logic by cognitive treating apparatus is carried out the identification/imagination realizes that the described identification/imagination is handled.It is in full accord that the logical and that described identification is handled is understood the logic of handling.Described identification process object can be an auditory information 107, visual information 108, and perhaps somatesthesia information 109.Shown in the logic handled of the imagination be to convert the new knowledge visual information to new sensory information.The grammar processing unit that at length introduce below in the epistemic logic treating apparatus of the present invention is carried out the process that grammer is handled.
Referring to Figure 16 a, described grammar processing unit is carried out general syntax (Universal Grammar) and is handled.Described general syntax is handled and is converted natural language to the unified form machine language, thereby realizes the formalization operation of machine.Described grammar processing unit comprises vocabulary processing subelement 173, and subelement 174 handled in phrase, and subelement 175 handled in the meaning of a word, and subelement 176, semantic processes subelement 177 and knowledge processing subelement 172 handled in sentence structure.Described grammar processing unit can according to language message import 170 declarative sentence/interrogative sentence classification or by the prompting of man-machine interface select to determine whether carrying out language message study storage still language ask for.If carry out the language learning process, described grammar processing unit judges whether that according to information needs carry out study of words again, phrase study, sentence structure study and lexical study etc.No matter be that process asked in study storing process or language, Language Processing all is to be delivered to senior level information processing layer by layer by elementary level information processing.
Referring to Figure 11 a, vocabulary is handled subelement 173 each vocabulary is carried out the vocabulary processing.When for example described cognition logic machine imported in " DOG " this speech, external information E (DOG) converted physical message PH (1000100,101111 in the machine to by ASCII compilation unit 131,1000111), convert perception location sign indicating number P/C (04,15,07) to by perception compilation unit 132 again.Vocabulary subelement 173 is interpreted as consciousness location sign indicating number C (103) from perception location sign indicating number P/C (04,15,07), and symbolic representation is P (04,15,07)＞C (103).Handle by same vocabulary, described cognition logic machine can obtain the consciousness location sign indicating number 102 of each vocabulary.Resulting vocabulary consciousness location sign indicating number 102 can convert perception location sign indicating number 104 again to.
Referring to Figure 11 b, to handle by same understanding, described phrase processing subelement 174 and semantic processes subelement 177 can be interpreted as the consciousness location sign indicating number 102 of phrase and statement from the perception location sign indicating number 104 of resultant vocabulary.For example " POODLE IS A KIND OF DOG " the words can be handled the consciousness location sign indicating number 102 that obtains this statement by same understanding, is formulated as P (101,102,222,103)＞C (301).
At described vocabulary, phrase, after the study of the meaning of a word and syntactic knowledge, described grammar processing unit 127 is according to vocabulary, phrase, the meaning of a word and syntactic knowledge are carried out semanteme generation 143.With reference to Figure 13, described basic semantic generative process 143 mainly is divided into three levels simultaneously.
First level is that level 141 asked in part of speech.After the attention memory of in short importing described epistemic logic processing unit, the understanding of epistemic logic processing unit/memory device carries out part of speech understanding/memory to this sentence to be handled, this processing procedure is shown among Figure 13 141, from the knowledge memory device, ask for out the part of speech 146 of this each word of statement, and it is deposited in the attention memory 114.For example the part of speech of each vocabulary of " POODLE IS A KIND OFDOG " the words can be understood/recall and is " NOUN; VERB; PHRASE; NOUN " by described cognition logic machine, be formulated as P (POODLE IS A KIND OFDOG)＞P/C (NOUN, VERB, PHRASE, NOUN)＜C (WOLD CLASS).
Second level is that level 142 asked in sentence pattern.The described part of speech of noting in the memory 146 that deposits in further carries out sentence/memory processing, and this sentence/memory processing procedure is asked for out corresponding sentence pattern 147 by the processing of sentence/memory shown among Figure 13 142, and deposits in the attention memory.Be formulated as
P(NOUN，VERB，PHRASE，NOUN)＞P/C((){}\/[])＜C(SYNTAX)。
Tri-layer is the semantic level 143 that generates.The described sentence pattern of noting in the memory 114 147 that deposits in generated the semantic structure 149 with syntactic structure with the statement that has before deposited in 148.The vocabulary of described study, phrase, the meaning of a word, sentence structure and semantic knowledge are stored in the knowledge memory device of the present invention.Be formulated as P (POODLE, IS, A KIND OF, DOG, (), { } ,/, []) → C ((POODLE) IS} A KIND OF/[DOG]).
Referring to Figure 16 b, if described grammar processing unit judges that language message is input as language and asks for process, the unified grammer that the language of 127 pairs of inputs of described grammar processing unit carries out from vocabulary to the statement is handled.For example question sentence " What is a kind of dog " at first understood/recalled for part of speech and phrase be expressed as P (What is a kind of dog)＞C/P (PRO, VERB, PHR, NOUN)＜C (WORDCLASS).Then the part of speech PRO that obtains, VERB, PHR, NOUN be understood that sentence pattern be expressed as P (PRO, VERB, PHR, NOUN)＞C/P (() { }/[])＜C (SYNTAX).Resulting sentence pattern P (() { }/[]) and the common generative semantics of the statement " What is a kind of dog " of input, be expressed as P (What is a kind of dog, () { }/[]) → C ((what) is} a kind of/[dog]).At last, the semanteme of generation is understood/is recalled for statement knowledge, is expressed as
P((what){is}\a?kind?of/[dog])＞C((poodle){is}\akind?of/[dog])。Wherein " poodle " is the object that " what " asked in the question sentence.Ask for the result and convert external information output to by perception compilation unit 132 described in Figure 10 and ASCII compilation unit 131.
Described grammar processing unit 127 can be a grade level tupe (Schema), be a series of from simple to complexity, from rudimentary to senior, the tupe that successively goes forward one by one, this tupe is the method that is called the level combination, the unit (element) that wherein rudimentary hierarchical information is senior hierarchical information.In like manner, to simply, the tupe that successively decomposes from senior level to rudimentary level is called the method for Hiberarchy Decomposition from complexity.The method of described combination and resolution process control is that the relational query of making up and resolution process converts between the unified machine address sign indicating number is handled, being about to all input informations converts to by elementary to senior machine address sign indicating number by described epistemic logic, wherein the primary address sign indicating number is the unit that forms senior address code, and the relation between primary address sign indicating number and the senior address code is set up and inquiry is handled realization by described epistemic logic.
This grade level tupe comprises vocabulary processing subpattern, and subpattern handled in phrase, and subpattern handled in the meaning of a word, and subpattern, semantic processes subpattern and knowledge processing subpattern etc. handled in sentence structure.It is prerequisite and the ingredient that next level is handled that each rudimentary level is handled.Described vocabulary tupe, the phrase tupe, meaning of a word tupe, the sentence structure tupe, whether semantic processes pattern and knowledge processing pattern are learnt/forget to handle or understanding/memory processing procedure according to inputting interface prompting decision.If learn/forget processing, described grammar processing unit is successively carried out study of words by above-mentioned study/forget processing procedure or is forgotten, phrase study or forget, lexical study or forget, sentence structure study or forget, semantic study or forget with knowledge learning or forget.If understanding/memory is handled, described grammar processing unit is successively carried out vocabulary understanding by understanding/memory processing procedure or is recalled, and phrase is understood or recalled, and the meaning of a word is understood or memory, and sentence structure is understood or recalled semantic understanding or memory and knowledge understanding or memory.Described grammar processing unit itself is not carried out direct computing, and it instructs epistemic logic processing unit 113 to carry out logical process.
The storing step of knowledge memory device:
Referring to Fig. 9, the knowledge memory device 115 that described epistemic logic information processing is concentrated is used for the relation between canned data and the information.Described knowledge memory device 115 is one and concerns tlv triple.The described tlv triple that concerns comprises sign indicating number territory, perception location 104, sensation sign indicating number territory, location 101 and sign indicating number territory, consciousness location 102.Sign indicating number territory, perception location 104 storage sensation location sign indicating number and consciousness location sign indicating numbers, sensation sign indicating number territory, location 101 storage sensation location sign indicating numbers, sign indicating number territory, consciousness location 102 storage consciousness location sign indicating numbers.Tuple (Tuple＜P/C, P, C 〉) the storage perception relation that concerns tlv triple.
Described knowledge memory device 115 can be deposited pairing various mapping relations in the above-mentioned relation triple store form according to the pre-assigned sequence of addresses of this system.Wherein, when in the storage of described knowledge memory device, storing, can be according to the requirement of system itself and to required canned data memory allocated address.Can store pairing all mapping relations according to other height of level of physical message in the machine during storage.Also can store pairing all mapping relations according to the priority of the order of external information input.
External information output step:
As Fig. 9 and 10, shown in, the information after output device 116 is used for system of the present invention handled outputs to coupled external unit.Epistemic logic of the present invention system converts perception location sign indicating number in machine physical message 130 by perception compilation unit 132 after handling knowledge information, by ASSCII compilation unit 131 physical message 130 in the described machine is converted to external information then and export.
Here it is different to want to be stressed that the present invention utilizes the method that cognition logic machine handles natural language and the recursive function arithmetic operation of traditional computer to handle, and it is the method that adopts relation storage and relational query.Described cognitive algorithm is intended to solve a series of problems that present natural language processing algorithm can't overcome, and comprises semantic singular problem, and the algorithm time is with the complexity index growing concern.
The cognitive disposal route of described natural language comprises the storage organization and the method for linguistry, and the general syntax rule of natural language.
The storage organization of described linguistry is different with traditional computer with method.At the traditional computer internal information is that one of instruction as data and service data connects a ground storage.The storage means of described knowledge is that the relation between information and the information is existed between the two memory structures of described grade.The storage means of described knowledge is storage instruction not.The all operations of described cognitive algorithm is stored in the epistemic logic processing unit, especially in grammar processing unit.The operation of traditional computer is the recursive function operation.The cognitive disposal route of described natural language is relation storage and relational query.
This method is to store and ask for by concern the sensory information in the tlv triple (Tuple＜P/C, P, C 〉) and the mapping relations of consciousness information that contain sensory information and consciousness information mapping relation to one.The described tlv triple that concerns comprises perception location sign indicating number, sensation location sign indicating number and consciousness location sign indicating number.The address that described sensation location sign indicating number is used to store sensory information.Described consciousness location sign indicating number is used to store the address of consciousness information.Described perception location sign indicating number is used to store the address of sensory information and the address of consciousness information simultaneously.The relation of one-to-many between storage sensory information and the consciousness information between described perception location sign indicating number and the consciousness location sign indicating number.The relation of one-to-many between storage consciousness information and the sensory information between described perception location sign indicating number and the sensation location sign indicating number.Based on the interchangeability of perception location sign indicating number, form the relation of multi-to-multi between described sensation location sign indicating number and the consciousness location sign indicating number by the conversion character of perception location sign indicating number.The physical message sign indicating number is one to one in described perception location sign indicating number and the machine.The physical message sign indicating number also is one to one with the information of extraneous input in the described machine.Described perception location sign indicating number generates according to physical message sign indicating number in the machine by the perception compilation unit.
Described concern the triple store form between can analogous language (comprising natural language and artificial language) information various relations and the processing procedure of relation.Described epistemic logic signal conditioning package concerns that to described sensory information and consciousness information in the triple store form learns/forget, and the understanding/memory and the identification/imagination are handled.Wherein the process of input information opening relationships triple store form can be finished by learning functionality by the epistemic logic system to external world.
The described triple store form that concerns can be carried out perception and be handled.Described perception handle be will input external information deposit knowledge memory in the form of perception location sign indicating number.
The described triple store form that concerns can be concluded processing.It is to set up mapping relations between sensory information and consciousness information that described conclusion is handled.
Describedly concern the processing of can deducing of triple store form, it is the process of asking for consciousness information mapping relation from sensory information that described deduction is handled.Concrete enforcement is in the mapping of the consciousness location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence.
The described triple store form that concerns can instead be unraveled silk processing, describedly instead unravels silk that to handle be process from consciousness requestor message sensory information mapping relations.Concrete enforcement is in the mapping of the sensation location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence.
The described triple store form that concerns can be carried out reasoning and be handled, and it is to ask for consciousness information from sensory information that described reasoning is handled, and converts sensory information then to, carries out the process of asking for of consciousness information again.
Described logical process process can adopt also, hands over, and is poor, changes, and four fundamental operations are operated and realized.
The practice significance that described perception is handled is with information conversion of equal value between state of feeling and consciousness state, is P=C with the perception formulate.For example P (ANIMAL)=C (ANIMAL) expression " ANIMAL " be sensory information also be consciousness information, can do conversion of equal value between them.
It is the relation of setting up between sensory information and the consciousness information that practice significance is handled in described conclusion, is P → C with the conclusion formulate.For example P (DOG) → C (ANIMAL) represents the relation of " DOG " and " ANIMAL ".
The practice significance that described deduction is handled is to occur simultaneously from sensory information to ask for consciousness information, is P＞C with the deduction formulate.For example P (DOG, CAT, MOUSE)＞C (ANIMAL) expression occurs simultaneously from sensory information " DOG, CAT, MOUSE " and ask for out consciousness information " ANIMAL ".The processing of deducing is a many reduction (≤m, many-one reducibility) computings.
The described counter practice significance of unraveling silk processing is to occur simultaneously from consciousness information to ask for sensory information, is P＜C with the anti-formulate of unraveling silk.For example P (CAT)＜C (PETANIMAL, CLIMB TREE) expression is asked for out sensory information " CAT " from consciousness information " PETANIMAL, CLIMB TREE " common factor.
The continuous process that described deduction is handled is that reasoning is handled, and is expressed as with rational formula
P (1)＞C (2)=P (2)＞C (3) is reduced to P (1)＞C (2)＞C (3).
For example P (POODLE)＞C (DOG)＞C (ANIMAL) expression " POODLE " is " DOG ", and " DOG " is " ANIMAL ".Instead unravel silk that to handle be many reduction operations.
It is the primitive rule of cognitive behavior that epistemic logic is handled.The external presentation of these logic rules is basic cognitive processes, comprises perception, learns/forget process, understanding/memory and identification/imagination process.In other words, the logical foundations of perception is the perception logic; The logical foundations of learning process is an inductive logic; The logical foundations of understanding and discerning is a deductive logic; The logical foundations of memory and imagination process is the anti-logic of unraveling silk.Between cognitive process and the logic rules one-to-one relationship is arranged.
Continuation is referring to Fig. 9, described perception by perception compilation unit 132 from extraneous 100 impression information and be stored as perception location sign indicating number 104.The meaning of described perception location sign indicating number 104 is that it is sensory information simultaneously, also is consciousness information, and it is corresponding one by one to form sensation and consciousness.For example
P (ANIMAL)=C (ANIMAL) expression " ANIMAL " be sensory information also be consciousness information, can do conversion of equal value between them.
Described learning process is set up relation between perception information and the consciousness information according to inductive logic, is information access process therefore.This process is an instinctive behavior for the people.For example the relation of sensory information " DOG " and consciousness information " ANIMAL " is learnt in P (DOG) → C (ANIMAL) expression.
Describedly forget the anti-process that process is a learning process, therefore the relation between its deletion perception information and the consciousness information is the delete procedure of information and information relationship.This process also is an instinctive behavior for the people.Be formulated as P%C.
Described understanding process concerns 102 according to the deductive logic 105 consciousness location sign indicating number of handling given perception location sign indicating number 104 that occurs simultaneously, for example, P (CAT, DOG, BIRD)＞C (ANIMAL) shows that " CAT, DOG, BIRD " can be understood as " ANIMAL ".
Described memory process concerns 101 according to the counter sensation location sign indicating number of unraveling silk the given perception location sign indicating number 104 of logical one 06 common factor processing, and for example: the capital of China is that Beijing can be expressed as P (Beijing)＜C (capital of China).Wherein " Beijing " is sensory information, and " capital of China " is consciousness information.
Because natural language processing of the present invention utilizes two memory relation storage of described sense-know grade and query manipulation, rather than recursive function arithmetic operation, realized the parallel processing of many-reduction, and overcome the instruction storage of present computing machine and function calculation method time index growing concern, thereby be applicable to the unified epistemic logic work of treatment of natural language information to natural language processing.
Above specific embodiment only be that some embodiments of the present invention are not to be used for limiting the present invention.Corresponding modify and change that those skilled in the art makes according to disclosed content all belong to the scope of protecting of the present invention.
Claims (22)
1. the cognition logic machine of anthropomorphic dummy's brain memory structure, it comprises input media, epistemic logic treating apparatus, knowledge memory device, output unit and the bus that is connected with above-mentioned each device;
Wherein said input media is used for external information is input to cognition logic machine;
Described epistemic logic treating apparatus changes into the information with certain storage format to the information from the input media input and stores into the knowledge memory device, described storage format is that described epistemic logic treating apparatus carries out logical process according to external information input to the mapping relations between sensory information and the consciousness information to the storage of sensory information and consciousness information mutual mapping relation;
Described knowledge memory device has transformed the logical process result that information with certain storage format and epistemic logic treating apparatus carry out and has stored above-mentioned;
Described output unit is used for the result output with the epistemic logic signal conditioning package.
2. cognition logic machine as claimed in claim 1, wherein said storage format is by concerning array or concerning that polynary group is realized.
3. cognition logic machine as claimed in claim 2, polynary group of wherein said relation are one to concern tlv triple Tuple＜P/C, P, C 〉, the described tlv triple that concerns comprises perception location sign indicating number P/C, sensation location sign indicating number P and consciousness location sign indicating number C;
The address that described sensation location sign indicating number is used to store sensory information, described consciousness location sign indicating number is used to store the address of consciousness information, the address that described perception location sign indicating number is used to store perception information simultaneously; Wherein
Between the sign indicating number of its corresponding sensation of described perception location sign indicating number location is the mapping relations of one-to-many;
Between the sign indicating number of its corresponding consciousness of described perception location sign indicating number location is the mapping relations of one-to-many;
By the sensory information of this perception location sign indicating number storage or consciousness information respectively and accordingly by the sensory information of sensation location sign indicating number storage or be the mapping relations of one-to-many between the consciousness information by the storage of consciousness location sign indicating number accordingly;
Described sensation location sign indicating number and and its corresponding consciousness location sign indicating number between conversion character by perception location sign indicating number form the mapping relations of multi-to-multi.
4. as each cognition logic machine in claim 1 or 3, wherein said epistemic logic treating apparatus also comprises the epistemic logic processing unit, notes mnemon and grammar processing unit;
The information that wherein said epistemic logic processing unit is used for importing is to external world carried out the cognition processing, and the result of this processing temporarily deposits needed input information when conduct is further handled in the attention mnemon under the guidance of grammar processing unit;
The infologic that described grammar processing unit provides general syntax rule control to instruct the epistemic logic processing unit is handled, and it is the vocabulary of natural language, phrase, and the meaning of a word, statement and syntactic structure convert unified machine language rule to;
Required information of using or result were temporarily stored when described attention mnemon was used for grammar processing unit and epistemic logic processing unit processes, the temporary information that should store can be by grammar processing unit simultaneously, and epistemic logic processing unit and connected other unit or device institute call.
5. cognition logic machine as claimed in claim 4, it further comprises the perception encoding unit wherein said epistemic logic processing unit, learns/forget device, understanding/memory device and identification/imagination device; Wherein
Described perception encoding unit is used for that the mapping relations between described sensory information and the consciousness information are carried out perception to be handled, and it is that the external information of importing is compiled perception location sign indicating number that this perception is handled;
Described study/forget device is used to learn/forget processing, and it is the process of setting up the mapping relations between sensory information and the consciousness information that described study is handled, and describedly forgets that to handle be the process of the mapping relations between deletion sensory information and the consciousness information;
Described understanding/memory device be used for to the mapping relations between described sensory information and the consciousness information deduce/instead unravel silk processing;
Described identification/imagination device is used for looking, listen and the sensory information and the mapping relations between the consciousness information of these extraneous shape informations of body deduce/instead unravel silk processing; Wherein
It is the process of asking for that is caused consciousness information by sensory information that described deduction is handled, and describedly instead unravels silk that to handle be the process of asking for that is caused sensory information by consciousness information.
6. cognition logic machine as claimed in claim 5, wherein said epistemic logic processing unit are comparers.
7. cognition logic machine as claimed in claim 4, wherein said grammar processing unit also comprise vocabulary processing subelement, and subelement handled in phrase, and subelement handled in the meaning of a word, and subelement, semantic processes subelement and knowledge processing subelement handled in sentence structure; Wherein above-mentioned subelement is controlled the epistemic logic processing unit respectively and is realized vocabulary, phrase, the meaning of a word, sentence structure, the processing of semanteme and knowledge.
8. cognition logic machine as claimed in claim 5, interior physical message of machine after the information that wherein said perception encoding unit is used for importing to external world transforms and perception location sign indicating number are encoded and are deciphered, and physical message is one to one in perception location sign indicating number that this compiled and the machine.
9. cognition logic machine as claimed in claim 8, wherein said perception location sign indicating number are automatically along the natural number of compiling according to the external information input sequence.
10. cognition logic machine as claimed in claim 4, wherein said grammar processing unit provide the control of general syntax rule, and this general syntax rule control comprises the vocabulary processing controls, the phrase processing controls, meaning of a word processing controls, sentence structure processing controls, semantic processes control and knowledge processing control.
11. cognition logic machine as claimed in claim 4, wherein said cognition logic machine also comprise an ASCII edit cell, it is used for the information that information with external world input is converted into physical message sign indicating number in the machine and physical message sign indicating number in the machine is converted into extraneous input.
12. as each cognition logic machine in claim 1 or 3, wherein this input media takes a kind of input or multiple input mode to import simultaneously, this input mode comprises the keyboard input, phonetic entry and figure input.
13. as each cognition logic machine in claim 1 or 3, wherein this output unit takes a kind of output or the multiple way of output to export simultaneously, this way of output is set according to the concrete entry terminal that is connected with described cognition logic machine.
14. epistemic logic information processing method of simulating human brain, this method is by any described cognition logic machine among the claim 1-13 each external information of importing to be converted it into the information with certain storage format, described storage format is the storage to sensory information and consciousness information mutual mapping relation, the epistemic logic treating apparatus of described cognition logic machine carries out logical process according to external information input to the mapping relations between sensory information and the consciousness information, the simulation human brain the process of information processing is realized study to the input external information, forget, conclude, deduce and instead unravel silk processing;
Described storage format is by concerning array or concerning that polynary group realizes;
Polynary group of described relation is one to concern tlv triple Tuple＜P/C, P, C 〉, the described tlv triple that concerns comprises perception location sign indicating number P/C, sensation location sign indicating number P and consciousness location sign indicating number C.
15. the epistemic logic information processing method of simulation human brain as claimed in claim 14, wherein above-mentioned perception location sign indicating number are automatically along the natural number of compiling according to the external information input sequence.
16. the epistemic logic information processing method of simulation human brain as claimed in claim 15, when wherein said information processing is the natural language information processing, employing concerns storage and the method for asking for to natural language information, this method is to concern tlv triple Tuple＜P/C by what contain sensory information and consciousness information mapping relation to one, P, C〉in sensory information and the mapping relations of consciousness information store, and utilize the general syntax rule of natural language to control, carry out the epistemic logic processing to having the language message that concerns the triple store form;
Wherein the operation of this epistemic logic processing is carried out by the epistemic logic processing unit, and described general syntax rale store is in grammar processing unit.
17. the epistemic logic information processing method of simulation human brain as claimed in claim 16, wherein said epistemic logic processing unit is handled the described epistemic logic that concerns that sensory information in the triple store form and consciousness information carries out, and comprises that perception, conclusion, deduction, counter unraveling silk with inference logic handle;
Described perception handle be will input external information deposit the knowledge memory device in the form of perception location sign indicating number;
It is to set up mapping relations between sensory information and consciousness information that described conclusion is handled;
It is the process of asking for consciousness information mapping relation from sensory information that described deduction is handled, and concrete disposal route is in the mapping of the consciousness location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence;
Describedly instead unravel silk that to handle be process from consciousness requestor message sensory information mapping relations, concrete disposal route is in the mapping of the sensation location sign indicating number processing of occuring simultaneously with the perception location sign indicating number of input information correspondence;
It is to ask for consciousness information from sensory information that described reasoning is handled, and converts sensory information then to, carries out the process of asking for of consciousness information again.
18. the epistemic logic information processing method of simulation human brain as claimed in claim 17, wherein said epistemic logic processing procedure adopts also, hands over, and is poor, changes four fundamental operations and operates and realize;
It is with information conversion of equal value between state of feeling and consciousness state that described perception is handled, and is P=C with the perception formulate, the perception processing be with same information stores in two addresses one to one, perhaps two of map addresses information one to one;
It is the relation of setting up between sensory information and the consciousness information that described conclusion is handled, and is P → C with the conclusion formulate;
It is to occur simultaneously from sensory information to ask for consciousness information that described deduction is handled, and is P＞C with the deduction formulate, and its expression is occured simultaneously from sensory information and asked for out consciousness information, and described deduction processing is the many reduction operations from perception location sign indicating number to consciousness location sign indicating number;
Describedly instead unraveling silk that to handle be to occur simultaneously from consciousness information to ask for sensory information, is P＜C with the anti-formulate of unraveling silk, and its expression is occured simultaneously from consciousness information and asked for out sensory information, and this is counter unravels silk that to handle be to many reduction operations of feeling the location sign indicating number from perception location sign indicating number;
P in the wherein above-mentioned formula represents to feel the location sign indicating number, and C represents consciousness location sign indicating number.
19. the epistemic logic information processing method of simulation human brain as claimed in claim 18, wherein said perception is concluded, and deduction and the anti-outside cognitive process of unraveling silk are to learn/forget, understanding/memory, and the identification/imagination and reasoning are handled;
Described learning process is perception and inductive logic information process, this process perception and inductive logic formulate;
Described understanding and identifying are the deductive logic processing procedures, this process deductive logic formulate;
The described memory and the imagination are the anti-logical process processes of unraveling silk, and this process represents that with the anti-logical formula of unraveling silk the memory here is meant the anti-process of unraveling silk of Word message, and the imagination is meant the anti-process of unraveling silk of acoustic information and graphical information;
Described reasoning process is the process that continuous deduction is handled, this process deduction formulate.
20. the epistemic logic information processing method of simulation human brain as claimed in claim 16, the process that wherein said grammar processing unit is handled be, at first judges according to the information of input and learn storing process or carry out language and ask for process;
If learn storing process then grammar processing unit need further to judge again be to carry out which kind of study, the study here comprises study of words, phrase study, lexical study and sentence structure study, after this learning process finishes, described grammar processing unit is to the vocabulary after learning, phrase, the meaning of a word and syntactic knowledge carry out semanteme and generate;
If described grammar processing unit judges that language message is input as language and asks for process, the language message of input is carried out the semanteme generation to described grammar processing unit and sentence comprehension/memory is handled, and convert the result to external information output by perception compilation unit and ASCII compilation unit.
21. the epistemic logic information processing method of simulation human brain as claimed in claim 20, wherein said semantic generative process mainly is divided into three levels, first level is that level asked in part of speech, grammatical category information corresponding to each independent word of forming this nature statement in the natural statement that will import in first level carries out the processing of part of speech understanding/memory by the understanding in the epistemic logic processing unit/memory device, asks for out from the knowledge memory device and the information of asking for is deposited in the attention mnemon;
Second level is that level asked in sentence pattern; In second level, to further carry out sentence/memory processing to depositing the part of speech of noting in the mnemon in the ground floor in by the understanding in the epistemic logic processing unit/memory device, from the knowledge memory device, ask for out corresponding sentence pattern, and deposit in the attention mnemon;
Tri-layer is the semantic level that generates; In tri-layer, in above-mentioned two levels, deposited the natural statement of above-mentioned input in note in the mnemon sentence pattern and part of speech respectively, carry out sentence comprehension/memory by the understanding in the epistemic logic processing unit/memory device and handle the semantic structure statement that generation has syntactic structure, and it is deposited in the knowledge memory device.
22. epistemic logic information processing method as claim 20 or 21 described simulation human brains, wherein the processing of grammar processing unit realizes by the control of general syntax rule, wherein this general syntax rule control is meant the vocabulary with natural language, phrase, the meaning of a word, statement and syntactic structure convert unified machine language rule to;
The control of described general syntax rule comprises the vocabulary processing controls, the phrase processing controls, meaning of a word processing controls, the sentence structure processing controls, semantic processes control and knowledge processing control, described grammar processing unit is the method that adopts level of information combination and resolution process control when carrying out above-mentioned control; The method of wherein said level combination is elementary level information processing, is delivered to senior level information processing method layer by layer; The method of described resolution process control is from senior level information processing, decomposes rudimentary level information processing method layer by layer;
The method of described combination and resolution process control is that the relational query of making up and resolution process converts between the unified machine address sign indicating number is handled.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2006100548172A CN1838159B (en) | 2006-02-14 | 2006-02-14 | Cognition logic machine and its information processing method |
| HK07103175.4A HK1097629A1 (en) | 2006-02-14 | 2007-03-23 | A cognitive logic machine and a method of information operation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2006100548172A CN1838159B (en) | 2006-02-14 | 2006-02-14 | Cognition logic machine and its information processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1838159A CN1838159A (en) | 2006-09-27 |
| CN1838159B true CN1838159B (en) | 2010-08-11 |
Family
ID=37015548
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2006100548172A Expired - Fee Related CN1838159B (en) | 2006-02-14 | 2006-02-14 | Cognition logic machine and its information processing method |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN1838159B (en) |
| HK (1) | HK1097629A1 (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101576445B (en) * | 2009-06-03 | 2010-12-01 | 重庆大学 | Data reappearing method for structure health monitoring failure sensor simulating memory of people |
| WO2014165752A1 (en) * | 2013-04-04 | 2014-10-09 | Sherwin Han | A polynomial method of constructing a non-deterministic (np) turing machine |
| US9418137B2 (en) * | 2014-08-29 | 2016-08-16 | Sherwin Han | Database without structured query language |
| CN104750828B (en) * | 2015-03-31 | 2018-01-12 | 克拉玛依红有软件有限责任公司 | A kind of unconscious self-learning method of conclusion deductive knowledge based on 6W rules |
| CN105304082B (en) * | 2015-09-08 | 2018-12-28 | 北京云知声信息技术有限公司 | A kind of speech output method and device |
| CN106897950B (en) * | 2017-01-16 | 2020-07-28 | 北京师范大学 | Adaptive learning system and method based on word cognitive state model |
| CN109683840A (en) * | 2018-12-25 | 2019-04-26 | 北京心法科技有限公司 | The exciting method and device of plyability cognitive information processing mechanism |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1078053A (en) * | 1992-04-27 | 1993-11-03 | 福建师范大学 | Injection electroluminescent nerve network circuit |
| CN1236138A (en) * | 1998-05-18 | 1999-11-24 | 中国科学院声学研究所 | Natural language statement analyzing method simulating brain's language sensing process |
| CN1700250A (en) * | 2004-05-17 | 2005-11-23 | 中国科学院半导体研究所 | Special purpose neural net computer system for pattern recognition and application method |
-
2006
- 2006-02-14 CN CN2006100548172A patent/CN1838159B/en not_active Expired - Fee Related
-
2007
- 2007-03-23 HK HK07103175.4A patent/HK1097629A1/en not_active IP Right Cessation
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1078053A (en) * | 1992-04-27 | 1993-11-03 | 福建师范大学 | Injection electroluminescent nerve network circuit |
| CN1236138A (en) * | 1998-05-18 | 1999-11-24 | 中国科学院声学研究所 | Natural language statement analyzing method simulating brain's language sensing process |
| CN1700250A (en) * | 2004-05-17 | 2005-11-23 | 中国科学院半导体研究所 | Special purpose neural net computer system for pattern recognition and application method |
Also Published As
| Publication number | Publication date |
|---|---|
| HK1097629A1 (en) | 2007-06-29 |
| CN1838159A (en) | 2006-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gregg | Knowledge and cognition | |
| CN1838159B (en) | Cognition logic machine and its information processing method | |
| Levinson et al. | Differential ineffability and the senses | |
| Turing | Mind | |
| Pyysiäinen | Magic, miracles, and religion: A scientist's perspective | |
| Collins et al. | Readings in cognitive science: A perspective from psychology and artificial intelligence | |
| CN108009285A (en) | Forest Ecology man-machine interaction method based on natural language processing | |
| Marciszewski et al. | Mechanization of reasoning in a historical perspective | |
| Mainzer | Artificial intelligence-When do machines take over? | |
| Reeves | Cognition and complexity: The cognitive science of managing complexity | |
| Phemister | The Rationalists: Descartes, Spinoza and Leibniz | |
| Kommers | Cognitive support for learning: imagining the unknown | |
| Cobley | Peirce in contemporary semiotics | |
| Jin | Relevance Between Artificial Intelligence and Cognitive Science | |
| Gregg | Rhetoric and knowing: The search for perspective | |
| Filson | Mind:" the Power of the Human Spirit" | |
| Ratsch et al. | Intelligence and artificial intelligence: an interdisciplinary debate | |
| Priss | Associative and formal concepts | |
| Yu et al. | Hot Wind, Cold Sun: Kuhn, Vygotsky, Halliday and Metaphors in Science and Science Education | |
| Aberšek et al. | Philosophical and Social Realm | |
| Kilic | Intelligent reasoning on natural language data: a non-axiomatic reasoning system approach | |
| Turing | Mind a Quarterly Review or Psychology and Philosophy Computing Machinery and Intelligence | |
| Magnani | External diagrammatization and iconic brain co-evolution | |
| O'Halloran | Mystifying discourse: a critique of current assumptions and an alternative framework for analysis | |
| Cai et al. | The beginning and evolution of consciousness |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1097629 Country of ref document: HK |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: GR Ref document number: 1097629 Country of ref document: HK |
|
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100811 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |