CN108255814A - The natural language production system and method for a kind of intelligent body - Google Patents
The natural language production system and method for a kind of intelligent body Download PDFInfo
- Publication number
- CN108255814A CN108255814A CN201810073361.7A CN201810073361A CN108255814A CN 108255814 A CN108255814 A CN 108255814A CN 201810073361 A CN201810073361 A CN 201810073361A CN 108255814 A CN108255814 A CN 108255814A
- Authority
- CN
- China
- Prior art keywords
- sentence
- intelligent body
- production system
- natural language
- clause
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9032—Query formulation
- G06F16/90332—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/30—Creation or generation of source code
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a kind of natural language production systems of intelligent body:Obtain nature language statement;Most simple thoughtcast clause set is obtained like predicate calculus form by natural sentence;It carries out production system and represents conversion;Carry out creative thinking study;Finally control intelligent body makes corresponding actions;Or generated statement output is as conclusion;Or it is stored as learning outcome.Existing artificial intelligence is proposed for subversive methodology, the method for switching to intelligent body thinking carrier completely by natural language word by human thinking is provided, so as to fulfill the consistency of man-machine thinking, realize the same thinking of intelligent body image people and with people's interaction.
Description
Technical field
The present invention relates to a kind of natural language word processor more particularly to a kind of natural languages applied to intelligent body
Production system.
Simultaneously the invention further relates to a kind of natural language literal processing method more particularly to it is a kind of applied to intelligent body from
Right language production method.
Background technology
With the fast development of artificial intelligence, various artificial intelligence products are obtained in all sectors of society widely should further
With.But people is directly exchanged with intelligent body using Human Natural Language, is the ultimate aim for realizing artificial intelligence.At present, due to
The shortcoming of methodology and technology path, artificial intelligence do not accomplish this point much still." the natural sentence that the present invention will refer to
Like predicate calculus form " in patent application number be:201610349629.6《A kind of natural language intelligent body recognition methods and it is
System》In have detailed statement.
Invention content
In order to solve the above-mentioned technical problem, present invention aims at provide a kind of natural language production system of intelligent body
And method.So that natural language word is decomposed into computer readable statement by intelligent body according to algorithm automatically, and carry out corresponding
Learn and make relevant action.
A kind of natural language production system of intelligent body of the present invention, which is characterized in that including:
Input unit, for obtaining nature language statement;
Cutting unit, for natural language sentence to be obtained most simple thoughtcast like predicate calculus form by natural sentence
Clause set;
Converting unit represents conversion for most simple thoughtcast clause set to be carried out production system;
Unit, for transformed sentence to be carried out creative thinking study;
Output unit, for intelligent body to be controlled to make corresponding actions;Or generated statement output is as conclusion;Or as study
As a result it is stored.
Preferably, the cutting unit is converted to different levels most simple thinking mould formed by three by sentence cutting
Formula sentence set.
Preferably, establish that sentence is bunched and sentence is bunched outer knowledge item in the cutting unit, using clause as pointer,
Bunch outer knowledge item to the work supplement retrieval of indefinite concept in sentence.
Preferably, the converting unit will most simple thoughtcast clause set using the equivalence relation in natural language sentence
It carries out production system and represents conversion, including:The production system of knowledge and relationship is represented, to expressing the production with reasoning
System representation, the production system table to learning to represent with the production system of planning, the new sentence of natural language and phrase generate
Show.
Preferably, the output unit carries out self-programing according to transformed sentence, for controlling intelligent body
Itself action.
The natural language production method of a kind of intelligent body of the present invention, which is characterized in that include the following steps:
S1, nature language statement is obtained;
S2, natural language sentence is obtained into most simple thoughtcast clause set by natural sentence like predicate calculus form;
S3, most simple thoughtcast clause set is carried out to production system expression conversion;
S4, transformed sentence is subjected to creative thinking study;
S5, control intelligent body make corresponding actions;Or generated statement output is as conclusion;Or it is deposited as learning outcome
Storage.
Preferably, the step S2 is converted to different levels most simple thoughtcast formed by three by sentence cutting
Clause gathers.
It is described, establish that sentence is bunched and sentence is bunched outer knowledge item in the step S2, using clause as pointer, in language
Sentence bunches outer knowledge item to the work supplement retrieval of indefinite concept.
Preferably, the step S3 is carried out most simple thoughtcast clause set using the equivalence relation in natural language sentence
Production system represents conversion, including:The production system of knowledge and relationship is represented, to expressing the production system with reasoning
It represents, the production system learnt with the expression of the production system of planning, the new sentence of natural language and phrase generation is represented.
Preferably, the step S5 carries out self-programing according to transformed sentence, for controlling intelligent body certainly
Body acts.
A kind of natural language production system of intelligent body of the present invention, the advantage is that, can be applicable to artificial intelligence
Each function of energy intelligent body.Including following several aspects:1. using natural language sentence like predicate calculus Xing Shi Zhuan Change as generate
After formula system representation, the automatic various functions for carrying out thinking are realized.Method migration of any natural language sentence by the present invention
Afterwards, the form for being directly entered program that intelligent body will be become can be calculated automatically.2. intelligent body directly uses the right Yu Yan ﹙ Wen Zi ﹚ of Zi
Control to oneself behavior.3. realize that intelligent body directly uses expressions of the right Yu Yan ﹙ Wen Zi ﹚ of Zi to knowledge and relationship.4. it realizes
Intelligent body is directly using the right Yu Yan ﹙ Wen Zi ﹚ of Zi to expressing the expression with reasoning.5. realize that intelligent body directly uses the right Yu Yan ﹙ of Zi
Expressions of the Wen Zi ﹚ to spatiality and decision tree.Traditional spatiality can be directly by the right Yu Yan ﹙ of Zi with Decision Tree Rule expression
Wen Zi ﹚ represent to replace.6. realize that intelligent body directly learns intelligent body using the right Yu Yan ﹙ Wen Zi ﹚ of Zi the expression with planning;It realizes
With natural language representing for state, action and target is carried out like predicate calculus form.
The present invention is directed to the above-mentioned deficiency of existing artificial intelligence, is proposed for existing artificial intelligence with subversive method
By the method for switching to intelligent body thinking carrier completely by natural language word by human thinking being provided, so as to fulfill man-machine think of
The consistency of dimension, realize the same thinking of intelligent body image people and with people's interaction.
Description of the drawings
Fig. 1 is the natural language production system of intelligent body of the present invention and the workflow schematic diagram of method;
Fig. 2 is the structure diagram of the natural language production system of intelligent body of the present invention.
Specific embodiment
As shown in Figure 1, 2, the natural language production system and method for a kind of intelligent body of the present invention, passes through first
Input unit obtains nature language statement, and most simple thoughtcast clause set is obtained like predicate calculus form by natural sentence;
Cutting unit carries out production system and represents conversion;Creative thinking study is carried out in unit;Finally in output unit control
Intelligent body processed makes corresponding actions;Or generated statement output is as conclusion;Or it is stored as learning outcome.It is proposed for existing
There is artificial intelligence that there is subversive methodology, provide and intelligent body thinking is switched to by natural language word completely by human thinking
The method of carrier, so as to fulfill the consistency of man-machine thinking, realize the same thinking of intelligent body image people and with people's interaction.
To each principle of work and power of the present invention, modular introduction discussion is carried out:
Equivalence relation in natural language sentence:The concept of " equivalence " is critically important.Language reflects human thinking, and the mankind think
The phenomenon that dimension then reflects the external world and relationship.
Sentence bunch composition linguistic context instantly under, situation of equal value has very much, from the point of view of philosophy, as long as cause and effect
Relationship, " because " are exactly equivalence with " fruit ".Restriction person and the person of being defined, actor and behavior and its bear actor etc. and
Of equal value.These belong to the implication relation of broad sense.It is important to note that only applying passerby and behavior recipient when behavior has reality
During the causality that matter influences, the two is just of equal value.For example, " inspection " is identical with " seeing " essence, all belong to virtual acting behavior word.
The behavior recipient of " inspection " is " WHAT ", it is believed that passerby No is applied in " WHAT " and behavior an equivalence relation, and only and " inspection
Look into " there is equivalence relation.
Important equivalence principle in same sentence:
Xia Dang under ﹙ Tong mono- ﹚ linguistic context, while be associated between existent, from common " presence " in same linguistic context in itself
For, all it is reciprocal causation, and part represents whole information.The incidence relation expression of " presence " performance:
(IFX, THEN Y)
=(IFY, THEN X)
For=(X=Y)/* when substituting calculating, the two is of equal value.*/
=X BE Y
=X Y
=(IF X, THEN X Y)/* (IFX, THEN Y)=X Y, when substituting calculating, the two is of equal value.*/
The basic mode of thinking computation is exactly " equivalence substitutes " operation.This is also mankind's reasoning from logic function.It is so-called " of equal value
The i.e. all causalities of replacement ", cause and effect is of equal value, and there is cause and effect etc. equivalence relation person can mutually be replaced in thinking
Generation.Such as
IF f(x),THEN f(y)
F (x)=f (y)
F (x), f (y) are arbitrary sentence or functional expression, of equal value in thinking computation, can be substituted for each other.
Using this equivalence relation, could be realized in natural language is calculated like the sentence of predicate form between different words
It directly replaces, so as to easily make inferences calculating.It is general formula below:
By WBW patterns (" WHAT BE WHAT " pattern)
=(determiner [x] noun (WHAT1), BE, determiner [z] noun (WHAT2))
=(determiner [x] noun (WHAT1)=determiner [z] noun (WHAT2))
The expression of determiner [x] noun (WHAT1)/* Chinese of=determiner [z] noun (WHAT2).English
Be determiner [x] noun (WHAT1) of determiner [z] noun (WHAT2) */
WDW patterns (" WHAT DO WHAT " pattern)
=(determiner [x] noun (WHAT), determiner [y] verb (DO), determiner [z] noun)
=(determiner [x] noun (WHAT)=(determiner [y] verb (DO), determiner [z]
noun))
=(determiner [x] noun (WHAT) (determiner [y] verb (DO)=determiner [z] noun))
WBWH patterns (" WHAT BE WHERE " pattern)
=(determiner [x] noun (WHAT), BE, determiner [z] noun (Where))
=(determiner [x] noun (WHAT)=determiner [z] noun (Where))
WBST patterns (" WHAT BE STATE " pattern)
=(determiner [x] noun (WHAT), BE, determiner [z] adjective (State))
=(determiner [x] noun (WHAT)=determiner [z] adjective (State))
Three in most simple thoughtcast are three functions.
When determiner is empty set, sentence is simplest formula.Determinant determiner ([x], [y], [z]) is
The ingredients such as matter, amount limit and description limits.
Most basic thoughtcast is four kinds, and natural language clause is just made of basic thoughtcast.General sentence is (here
The sentence of finger is the statement part using punctuation mark as boundary) will not be too long, main body frame is formed with single-mode in whole sentence level
Sentence be more, do not exceed three basic models at most.An it will be understood that language for including more than two basic models
Recurrence form is presented in sentence, sentence structure.Natural language clause is made of basic thoughtcast.It is calculated according to permutation and combination, shares 28
The clause of kind various combination.
The application for a patent for invention project that natural language has been previously proposed like predicate calculus form in the applicant《A kind of nature
The recognition methods of language intelligence body and system》In have detailed statement.The application for a patent for invention number or Patent No.:
201610349629.6。
Using equivalence relation, sentence is bunched arbitrary reasoning of the natural language like the dissenting words in predicate calculus form.
Sentence is bunched like all information that former natural language sentence is fully retained in predicate calculus form, original cause and effect
Relationship is all hidden in the basic thoughtcast clause like predicate calculus form (until most simple thoughtcast clause set).One complete
During whole sentence or sentence are bunched, descriptor WHAT and DO even can be equivalent to sentence or sentence is bunched.After same sentence segmentation,
Clause between different level is of equal value.Three (it is extra that middle entry BE is also believed to) in basic thoughtcast are individually etc.
Valency.Equivalence relation using between the Components (word) of basic thoughtcast and Components group (phrase) is as substituting operation
Tool, can be achieved with sentence and bunch like the arbitrary reasoning of the dissenting words in predicate calculus form.
This is the major way of human thinking's reasoning, by it is between items in thinking basic model in linguistic context instantly, limited
The equivalence relation for determining between vocabulary and restriction vocabulary and to be defined between clause and restrictive clause is substituted and is obtained different
The arbitrary of word transmits lead, so as to fulfill reasoning and its conclusion.
See the following concrete analysis bunched to sentence:
Input natural language sentence:" there is well landscape outside window in the beautiful house of Lao Wang, is one before room in lakeside
Small garden.There is very high value in the house for having good landscape, and subsidiary valuable thing can more improve its value." therefore, it can
To infer:
There is very high value in the house of Lao Wang;
Subsidiary valuable small garden can more improve the value in the house of Lao Wang.
The beautiful house of Lao Wang is in lakeside
=(house of Lao Wang, in lakeside)
=(house ((Lao Wang, HAS, house) ∧ (house, IS are beautiful)), in lakeside)
It obtains
(house, in lakeside) ∧ (Lao Wang, HAS, house) ∧ (house, IS are beautiful)
Lao Wang=Lao Wang has
(house, in lakeside)=house is in lakeside=lakeside house
The house of (Lao Wang, HAS, house)=Lao Wang
(house, IS are beautiful)=house is beautiful
Be defined clause (Lao Wang, HAS, house)=restrictive clause (house, IS are beautiful)=restrictive clause (house,
In lakeside), i.e.,
House=house of Lao Wang is beautiful=lakeside house
(house of Lao Wang) has well landscape outside window
=((Lao Wang, HAS, house) house, has, well outside window landscape)
=(house has, well outside window landscape) ∧ (Lao Wang, HAS, house)
=(house has, landscape (landscape, BE, ∧ is good outside window) ∧ (Lao Wang, HAS, house))
=(house has, landscape (landscape, BE, outside window ∧ good (good, BE, very)) ∧ (Lao Wang, HAS, house))
=(house has, good landscape (landscape, BE, outside window ∧ (good, BE, very)) ∧ (Lao Wang, HAS, house))
=(house has, good landscape) ∧ (Lao Wang, HAS, house) ∧ (landscape, BE, outside window ∧ (good, BE, very))
By (house has, good landscape) ∧ (Lao Wang, HAS, house)
=(Lao Wang, HAS, house (house has, good landscape))/* is defined clause (Lao Wang, HAS, house)=restriction
Sentence (house has, good landscape).I.e.:(house of Lao Wang)=(house for having good landscape) */
=(Lao Wang, HAS, (house has, good landscape))
=(Lao Wang, HAS, (house for having good landscape))
=((house for having good landscape) of Lao Wang)
The house for having good landscape of=Lao Wang
=there is the house of good landscape
It is a small garden before room
=(before room, is, a small garden)
=(In front of the house=garden a small garden mono- small before room)
A small garden before=room
Before room=house before In front of the house
There is front in=house (gathering around)
=front belongs to (in) house
There are one small gardens before room
=(before room, having, a small garden)
Before=(before (house has, preceding), having, a small garden)/*=front.*/
=(house has, preceding) ∧ (preceding, to have, a small garden)
=(house has, a small garden) ∧ (preceding, to have, a small garden)
=(house has, small garden) ∧ (preceding, to have, small garden) ∧ (small garden, BE, one)
It is that there are one small garden → house, there are one little Hua before one small garden → room before one small garden → room before room
Garden
There is very high value in the house for having good landscape
=(having the house of good landscape, have, very high value)
=(house for having good landscape of Lao Wang, have, very high value)
There is very high value in the house for having good landscape of=Lao Wang
=(house of Lao Wang, has, very high value)
There is very high value in the house of=Lao Wang
There is the high value of the house of good landscape=very
Note:The house for having good landscape of Lao Wang
The house ∧ of=Lao Wang has the house of good landscape
=(house, BE, Lao Wang's) ∧ (house, BE have good landscape)
=(Lao Wang, HAS, house) ∧ (house has, good landscape)
It obtains
(Lao Wang, HAS, house)=(house has, good landscape)
The house of the good landscape in the house of Lao Wang=have
And subsidiary valuable thing can more improve its value
=and (subsidiary valuable thing, can more improve, value)
=∧ Add ((thing, BE, subsidiary ∧ are valuable) thing, (improve, BE, more can) improve, have good landscape
The value in house)
=∧ Add (thing improves, and has the value in the house of good landscape) ∧ (improve, BE, more can) ← (thing, BE are attached
The ∧ of band is valuable)
Again by " and subsidiary valuable thing can more improve its value " as pointer, to related notion " it is subsidiary,
Valuable, thing " etc. handles accordingly:
Search statement bunches outer knowledge item to supplement its deficiency.
WHAT1HAS WHAT2 represent that WHAT2 is that the adjunct of WHAT1 or WHAT1 are attached to WHAT2.
WHAT is the element in concept " thing ", i.e.,
WHAT=things, i.e.,
Small garden=thing
WHAT1HAS WHAT2
WHAT2 subsidiary=WHAT1
=(WHAT2, BE, WHAT1 subsidiary X (thing))/* supplement sentences are bunched outer knowledge, and (thing) is concept " east
The set in west ".*/
Again, it is valuable to obtain " small garden " for search knowledge base.I.e.
Small garden=valuable thing, by
(house has, small garden)
=(the subsidiary small garden in house)
=(the subsidiary valuable thing in house)
=(thing, BE, the ∧ that the ∧ in house is subsidiary are valuable)
=(thing, BE, house) ∧ (thing, BE, subsidiary ∧ are valuable)
=(small garden, BE, house) ∧ (small garden, BE, subsidiary ∧ are valuable)
It searches for and finds out Equivalent Form, its correlation is substituted into:
(thing improves, and has the value in the house of good landscape) ∧ (improve, BE, more can) ← (thing, BE, subsidiary ∧ have
Value)
=(small garden, improve, the value in the house of Lao Wang) ∧ (improve, BE, more can) ← (small garden, BE, subsidiary ∧
It is valuable)
=(small garden, (and improve, BE, more can) improve, the value in the house of Lao Wang) ← (small garden, BE, subsidiary ∧ have
Value)
=(small garden can more improve, the value in the house of Lao Wang) ← (small garden, BE, subsidiary ∧ are valuable)
=(small garden (small garden, BE, subsidiary ∧ are valuable), can more improve, the value in the house of Lao Wang)
=(subsidiary valuable small garden, can more improve, the value in the house of Lao Wang)
=subsidiary valuable small garden can more improve the value in the house of Lao Wang
It is to bunch itself to add by sentence to bunch outer knowledge with two supplement sentences and the reasoning completed above.
The meaning of one word is obtained by the e-learning that sentence is bunched using equivalence principle
By (x1 | [x])=(g1 | [g]),
[g] is Gestalt structural dimension collection, and g1 is the element that dimension is concentrated;[x] is genus collection.X1 is concept set
In an element.
Intelligent body learns:IF " g1 " and " x1 " are of equal value in a linguistic context, are partly to represent entirety." g1 " is " x1 "
One of denominator format tower structure dimension, i.e., both matching, g1 is also a metanotion,
The two has the same or similar Gestalt structural dimension.If g1 it is known that can determine unknown word x1 and g1 accordingly
There is the overlapping of Gestalt structural dimension.If matching is enough to quantity, reaches whole overlappings, then can determine this unknown word x1's
Full format tower structure dimension, so that it is determined that its meaning of a word.The meaning of a word determines by its Gestalt structural dimension set, therefore,
IF " (g1, g2 ..., gj | [g]) " and " x1 " equivalence in j linguistic context, it is that all parts represent entirety.“gj”
One of the denominator format tower structure dimension of " x1 ", i.e. the two matching, gj is also a metanotion,
THEN x1 are by (g1, g2 ..., gj | [g]) definition.
One a block of ice=cold of example
By a block of ice=cold
IF " ice " and " cold " are of equal value in this linguistic context, are partly to represent entirety." cold " is the propcrty lattice of " ice "
One of formula tower structure dimension, i.e. the two matching, ice contains cold(cold or metanotion)
G1=cold
The two has the same or similar Gestalt structural dimension.If a word it is known that can determine that is unknown accordingly
Word and known words have the Gestalt structural dimension Chong Die.If matching is enough to quantity, reaches whole overlappings, then can determine this
The full format tower structure dimension (g1, g2 ..., gj | [g]) of unknown word, so that it is determined that its meaning of a word.
Again by a block of ice=ice or by ice is a block of, have
Ice=a block of
One of two characterization Gestalt structural dimensions of the mat woven of fine bamboo strips of ice are obtained,
G2=a block of
Easily solid easyization of g3=
By matching of the word in simplest formula from different known words, its representation of word meaning is obtained.Intelligent body learns
One of function.This is related to the constructive method of the meaning of a word.
Such as upper example.
By " clearing up the misunderstanding ", there is equivalence relation:
Ice=releases previous ill will, and abstract expression is
Inverse state before ice=changes
Ice=is released
Ice=previous ill wills
Pay attention to:A principle was once referred to above:In linguistic context instantly, actor and behavior and its result are of equal value.
Such as " I is worker ", have
I=worker
" I goes to Beijing "
I=I go to Beijing
Go to Beijing=I go to Beijing
I=go to Beijing
I=it goes
Go to=Beijing
If it asks:Who goes to Beijing
Who goes to Beijing i.e.
Who=go to Beijing
By " I " replacement " who ".
Generate new sentence " I goes to Beijing ".
It is above-mentioned by natural language sentence by natural sentence like predicate calculus form, be converted to by different level by sentence cutting
Gathered by three most simple thoughtcast clauses formed;Establish that sentence is bunched and sentence is bunched outer knowledge item, using clause as
Pointer is bunched outer knowledge item to the work supplement retrieval of indefinite concept in sentence, is completed by cutting unit.
Illustrate that most simple thoughtcast clause set is carried out production system and represents the work converted by converting unit in detail below
Principle and the course of work:
The production system of natural language sentence represents
Traditionally, production is the logic rules of referred to as inference rule (inference rule), these rule allow from
New proposition is formed in old proposition.In the present invention:
Production system is one of simple expression-form of nature statement fuction.IF X THEN Y, regular meaning are:
IF X are true, and THEN Y are true.
One production system includes an ordering rule sequence, referred to as production rule (production rule) or
Production (production).Each rule writing " ci → ai ", wherein ci is condition part (condition part), and ai
It is effect part (action part).The two existing cause and effect each other in linguistic context instantly.Existing causality is used as,
Ai is condition part again, and ci is effect part.
One production system is rule set.Sentence, sentence Ju Qun ﹙ Duan Luo ﹚, Yu Pian ﹙ Zheng Pian ﹚ are exactly rule set.
c1→a1
c2→a2
……
ci→ai
……
cm→am
In general, a regular condition part may be the feature for handling the perception that sensor inputs and generating
Any binary value (0,1) function.
Since defined terms " BE " both sides of definition statement WHAT1BE WHAT2 are of equal value, so, the two reciprocal causation,
It can turn the production system rule that Change is IF THEN, so as to realize that automated reasoning, association, generation extended meaning are new under this form
The thought processes such as sentence:
In linguistic context instantly:
By definition statement WHAT1BE WHAT2, have
IF WHAT1BE WHAT2, THEN (WHAT1, BE, WHAT2)
IF (WHAT1, BE, WHAT2), THEN (IF WHAT2, THEN What1)
IF WHAT2, THEN What1.
What1 is WHAT1 or its variant.
Equally, in linguistic context instantly:
By WHAT1DO WHAT2, have
IF WHAT1DO WHAT2, THEN (WHAT1, DO, WHAT2)
IF (WHAT1, DO, WHAT2), THEN (IF WHAT1, THEN DO What2)
IF WHAT1, THEN DO What2.
What2 is WHAT2 or its variant.
By WHAT1DO WHAT2, have
IF DO WHAT2, THEN What1.
What1 is WHAT1 or its variant.
By DO WHAT2, have
IF DO WHAT2THEN (IF DO, THEN What2)
IF DO THEN What2
What2 is WHAT2 or its variant.
Equally, in linguistic context instantly:
By WHAT BE WHERE, have
IF WHAT, THEN WHERE.
By WHAT BE WHERE, have
IF WHERE, THEN WHAT.
Equally, in linguistic context instantly:
By WHAT BE STATE, have
IF WHAT, THEN STATE.
By WHAT BE STATE, have
IF STATE, THEN WHAT. etc..
The natural language production system of intelligent body
Natural language is like the algorithm that predicate calculus form Yu Ju Zhuan Change are that production system represents:Procedure nature languages
Say that the production system of clause WHAT1DO WHAT2 represents input:The natural language of WHAT1DO WHAT2 is like predicate calculus shape
Formula
output:The IF/THEN production system Rule Expressions of WHAT1DO WHAT2
It is that the algorithm that production system represents is similar that its basic thoughtcast of excess-three kind, which turns Change,.
Represent that natural language sentence segmentation is the process like predicate calculus form with production system:
Natural language is like predicate calculus form:
Every son of my father is all my brother
=(every son of my father, be all my brother)
=((my father, HAS, per son) be all per son, brother (I, HAS, brother))
=((father (father, HAS, I), HAS, per son) per son, be all, brother (I, HAS, brother))
=(per son, being all brother) ∧ (father, HAS, per son) ∧ (father, HAS, I) ∧ (I, HAS,
Brother)
=((son is brother) ∧ (father, HAS, son)) ∧ (father, HAS, I) ∧ (I, HAS, brother)
By (father, HAS, I) (I is son) of equal value, (son is brother) is substituted into, is had
(I, be, brother) or by (I, HAS, brother) (I is brother) of equal value in knowledge data base, draw (I,
It is brother).
With IF the production system rule of THEN sentence segmentation process above is expressed as follows:
(1) every son of my fathers of IF is all my brother, THEN (every son of my father, be all my brother)
Every son of my fathers of IF, THEN my brothers
(2) IF (my father, HAS, per son) every son, THEN brothers (I, HAS, brother)
IF is per son, THEN (my father, HAS, per son)
IF brothers, THEN (I, HAS, brother)
IF (I, HAS, brother), THEN my brothers/* notes:Equivalence relation in sentence.*/
(3) father IF, THEN (HAS, per son)
IF HAS, THEN are per son
Son IF, THEN are each
(4) father IF, THEN (HAS, I)
IF HAS, THEN I
IF I, THEN (HAS, brother)
IF HAS, THEN brother
Natural sentence like predicate calculus Xing Shi Zhuan Change be IF THEN production system Rule Expression:
There is no knowing where he has gone.
He has gone where not knowing.
There is no knowing where he has gone
=(There, is, no knowing where he has gone)
=(where he has gone, is, no knowing)
=(where (where, BE, he has gone), is, no knowing)
=(where, is, no knowing) ∧ (where, BE, he (he, has gone))
=(where, is, no knowing) ∧ (where, BE, he) ∧ (he, has gone)
Similar sentence:There is no knowing what he has done
What he, which has done, to know.
There is no knowing what he has done
=(what, is, no knowing) ∧ (what, BE, he) ∧ (he, has done)
General formula:(where, is, no knowing) ∧ (where, BE, he) ∧ (he, has gone)
=There BE what1what2what3HAS DONE
=(what2, BE, what1) ∧ (what2, BE, what3) ∧ (what3, HAS DONE)
BE and HAS determines odd or even number, present tense or perfect tense etc. by actor's part of speech;DONE is behavior word DO
Completion state.
IF THEN production system regular expression:
1. IF where, THEN no knowing
IF no knowing, THEN where
2. IF where, THEN he
IF he, THEN where
3. IF he, THEN has gone
IF has gone, THEN he
General formula:There BE what1what2what3HAS DONE specifically gather member with general word set symbolic
Plain vocabulary, with IF THEN production system regular expression:
1. IF what2, THEN what1
IF what1, THEN what2
2. IF what2, THEN what3
IF what3, THEN what2
3. IF what3, THEN HAS DONE
IF HAS DONE, THEN what3
The production system Rule Expression of sentence generating process
The subject core word past participle of if statement makees attribute, then since this attribute and word that it is modified are passive
Relationship is equivalent to the attributive clause of a passive voice.
Have
Most of the people invited to the party were famous scientists. and Most
of the artists invited to the party were from South Africa.
Therefrom automatically generate the sentence of new correlation justice.
①Most of the people invited to the party were famous scientists
=(Most of the people invited to the party, were, famous scientists)
=((Most of the people, IS, invited to the party), were, famous
scientists)
=(Most of the people, IS, invited to the party) ∧ (Most of the people,
Were, famous scientists)
With IF THEN production system regular expression:
IF Most of the people, THEN invited to the party
IF Most of the people, THEN famous scientists
②Most of the artists invited to the party were from South Africa.
=((Most of the artists, IS, invited to the party), were, from South
Africa)
=(Most of the artists, IS, invited to the party) ∧ (Most of the
Artists, were, from South Africa)/* pays attention to:IS is newly increased, clear in order to express, and can remove;And were
It is then original, is kept for generation newspeak sentence.*/
With IF THEN production system Rule Expression:
IF Most of the artists, THEN invited to the party
IF Most of the artists, THEN (were, from South Africa)
IF were, THEN from South Africa
IF from, THEN South Africa
Two the first kernel sentence of sentence (Most of the X, IS, invited to the party) sentence patterns are identical, by
In artist ∈ people (artist belongs to people generics), so, may be can be used to replace were, and use people
Artists in replacing sentence 2. replaces most with some, automatically generates
(3. Some of the people, BE, invited to the party) ∧ (Some of the people,
May be, from South Africa)
With IF THEN production system Rule Expression:
IF the people, THEN the artists/*artist belong to people generics,
Artist ︱ ﹝ people ﹞ — — ﹝ people ﹞ are the collection of people, and artist is one of element.
This is general rule.*/
IF Most, THEN some/*Most contain some, and some ∈ Most, this is general rule.*/
IF Most of the people, THEN Most of the artists
IF (some is substituted, Most), THEN (may be are substituted, were)
IF were, THEN may be
IF Most of the people, THEN Some of the people
IF Some of the people, THEN invited to the party
IF Some of the people, THEN (may be, from South Africa)
IF may be, THEN from South Africa
IF from, THEN South Africa
Backtracking generation sentence
Some of the people invited to the party may be from South Africa.
Output.
With IF THEN production system Rule Expression:
Since preceding paragraph IF is identical, merge following two IF THEN production system rule 1. and 2., be expressed as 3.:
1. IF Some of the people, THEN invited to the party
2. IF Some of the people, THEN (may be, from South Africa)
3. IF Some of the people, THEN invited to the party ∧ (may be, from South
Africa)
Cancel " IF " " THEN " " ∧ " ", ", complete backtracking, generate nature sentence:Some of the people
invited to the party may be from South Africa.
With production system Rule Expression nature or social principle:
Social unit and whole general rule:
If necessary pay is the agreement without constraining really for overall goals, a Social Individual wishes itself not
It pays and can benefit simultaneously from overall goals, then he can be avoided that undertaking each individual corresponding with overall goals realization answers
When what is undertaken pays.If each Social Individual takes the behavior opposite with overall goals, the effect of individual behavior summation
It is antipodal with overall goals.
When individual amount is numerous, perhaps an individual can be guessed ﹙, itself take the behavior opposite with overall goals, because
For individual amount it is numerous without influence overall goals realization, then he implement this behavior.His idea is also likely to be
Other most or all individuals conjectures and behavior, then said circumstances will be realized.﹚
If x collects for individual, P is overall goals, and u is beneficial concept set,
(P, BE, useful for all x) --- " for all x, P is beneficial u ".
With this sentence pattern of IF/THEN production systems Rule Expression:
1. IF P, THEN (useful, for, all x)
2. IF P, THEN useful
3. IF P, THEN (for, all x)
4. IF useful, THEN (for, all x)
(x, freely give to, P) --- " for P, all x take paying for non-supervised (strong) restraining force
With this sentence pattern of IF/THEN production systems Rule Expression:
1. IF all x, THEN (freely give to, P)
2. IF freely give, THEN (to, P)
3. IF give, THEN freely
4. IF to, THEN P
(x, same give for, P) → (all x, obtain from, P) --- " for P, all x take identical
It paysU " then can be benefited from P to all x.
With this sentence pattern of IF/THEN production systems Rule Expression:
1. IF all x, THEN (same give for, P)
2. IF same give, THEN (for, P)
3. IF give, THEN same
4. IF for, THEN P
5. IF (all x, same give for, P), THEN (all x, obtain from, P)
6. IF all x, THEN (obtain from, P)
7. IF obtain, THEN (from, P)
8. IF from, THEN P
(y, not same give for, P) → (x, like, y) --- it " for P, is not taken if there is a y identical
It paysIt, all may be as y then to all x ".
With this sentence pattern of IF/THEN production systems Rule Expression:
1. IF y, THEN (not same give for, P)
2. IF not same give, THEN (for, P)
3. IF give, THEN not same
4. IF for, THEN P
5. IF (y, not same give for, P), THEN (all x, like, y)
6. IF all x, THEN (like, y)
7. IF like, THEN y
(all x, like, y) → (all x, not obtain from, P) --- " for all x, all as y, then
To all x, overall goals P " cannot be all obtained.
With this sentence pattern of IF/THEN production systems Rule Expression:
1. IF all x, THEN (like, y)
2. IF like, THEN y
3. IF (all x, like, y), THEN (all x, not obtain from, P)
4. IF all x, THEN (not obtain from, P)
5. IF not obtain, THEN (from, P)
6. IF from, THEN P
Note:X, y=What, i.e. variable x, y may be single concept, it is also possible to one statement, etc..
The member that beneficial u is concentrated is known as " good, profit, benefit are used, and harvest obtains " etc..
The production system of natural sentence reasoning represents:
Natural sentence is:
Before first, second, third and fourth are raced, they respectively predict result:
First prediction second will win;
Second prediction fourth will be whipper-in;
Third prediction first is third;
The prediction of fourth prediction first will be correct.
These predictions only there are one being correct, and be the prediction that last victor makes, provide first, second, third,
The name minor sort of fourth race.
Preceding paragraphs quoted from《Computer science outline》10th edition, J.Glenn Brookshear write, p1, p140.
First prediction second will win
=(first, prediction, second will win)
=(first, prediction, (second, it will obtain, win))
=(first, prediction, second) ∧ (second, it will obtain, win)
(1) IF (first, prediction, second), THEN (second, it will obtain, win)
IF first, THEN (prediction, second (second, it will obtain, win))
IF predictions, THEN second (second, it will obtain, win)
IF predictions, THEN (second, it will obtain, win)
IF predictions, THEN (second, it will BE, first place)
IF first, THEN (IF second, THEN first places)
IF predictions, THEN second
IF second, THEN first places
(second, it will obtain, win)=(second, it will BE, first place)
Second prediction fourth will be whipper-in
=(second, prediction, fourth will be whipper-in))
=(second, prediction, (fourth will be, whipper-in))
=(second, prediction, fourth) ∧ (fourth will be, whipper-in)
(2) IF (second, prediction, fourth), THEN (fourth will be, whipper-in)
IF second, THEN (prediction, fourth)
IF second, THEN (prediction, fourth (fourth will be, whipper-in))
IF predictions, THEN fourths (fourth will be, whipper-in)
IF fourths, THEN (fourth will be, whipper-in)
IF fourths, THEN (will be, whipper-in)
IF fourths, THEN will be
IF will be THEN whipper-ins
IF fourths, THEN whipper-ins
Third prediction first is third
=(third, prediction, first is third))
=(third, prediction, (first is third))
(3) IF (third, predict, first), THEN (first is third)
IF third, THEN (prediction, first (first is third))
IF predictions, THEN first (first is third)
IF first, THEN (first is third)
IF first, THEN thirds
The prediction of fourth prediction first will be correct
=(fourth, prediction, the prediction of first will be correct))
=(fourth, prediction, (prediction of first will be, correct))
=(fourth, prediction, (first, HAS, prediction) ∧ (prediction will be, correct))
=(fourth, prediction, first) ∧ | (first, HAS, prediction) ∧ (prediction will be, correct) |
=(fourth, prediction, first) ∧ | (first, HAS, prediction) ∧ (prediction will be, correct) ∧ (first will be, correct) |
(4) (first will be, correctly IF (fourth, prediction, first), THEN (first, HAS, prediction) ∧ (prediction will be, correctly) ∧
)
IF fourths, (prediction, ((first will be, correctly (first, HAS, prediction) ∧ (prediction will be, correctly) ∧ first THEN
)))
IF predictions, THEN first ((first, HAS, prediction) ∧ (prediction will be, correct) ∧ (first will be, correct))
IF fourths, THEN (first, HAS, prediction) ∧ (prediction will be, correct) ∧ (first will be, correct)
IF fourths, THEN (first, HAS, prediction)
IF fourths, THEN first
IF fourths, THEN (IF second, THEN first places)
IF (first, HAS, prediction), THEN (prediction will be, correct)
IF (first, HAS, prediction), THEN (first will be, correct)
IF first, THEN predictions
IF first, THEN (prediction will be, correct)
IF predictions, THEN (first will be, correct)
IF first, THEN (will be, correctly)
IF first, THEN ((second, it will obtain, win), will be, correctly)
IF ((second, it will obtain, win), will be, correctly), THEN ((IF second, THEN first places) will be, correct)
The prediction that victor makes is correct
=(prediction that victor makes is, correct))
=((victor makes, prediction), be, correctly)
=(victor makes, prediction) ∧ (prediction is, correct)
=(victor makes, prediction) ∧ (prediction is, correct) ∧ (victor is, correct)
(last victor=victor)
(5) IF (victor makes, prediction), THEN (prediction is, correct) ∧ (victor is, correct)
IF (victor makes, prediction), THEN (first place is made, prediction)
IF first places, THEN (make, predict)
IF first places, THEN predictions (prediction is, correct) ∧ (first place is, correct)
IF predicts that THEN is correct
IF first places, THEN are correct
Reasoning is as follows:
(i) all possible situation is first listed
IF first ∨ second ∨ the third ∨ fourths, THEN first places
IF first ∨ second ∨ the third ∨ fourths, THEN second places
IF first ∨ second ∨ the third ∨ fourths, THEN thirds
IF first ∨ second ∨ the third ∨ fourths, THEN fourths
(ii) replacement calculating is carried out by logical relation above
IF first ∧ fourths, THEN (IF second, THEN first places)
IF first ∧ fourths, THEN not first places
IF second, THEN not first places
IF (IF second, THEN first places), THEN (IF fourths, THEN whipper-ins)
IF (IF second, THEN not first places), THEN (IF fourths, THEN not fourths)
IF first ∧ second ∧ fourths, THEN not first places
IF first ∨ second ∨ the third ∨ fourths, THEN first places
IF third, THEN first place
IF (IF third, THEN first place), THEN (first, THEN thirds)
IF first, THEN thirds/* the third predict that first is third.*/
IF (IF fourths, THEN not fourths) ∧ (IF first, THEN thirds) ∧ (IF third, THEN first place), THEN (IF
Fourth, THEN second places) ∧ (IF second, THEN fourths)
(iii) it according to the order of names listed on a roster arranges
IF third, THEN first place/* backtracking generated statements:Third is first place.*/
IF fourths, THEN second places/* backtracking generated statements:Fourth is second place.*/
IF first, THEN thirds/* backtracking generated statements:First is third.*/
IF second, THEN fourths/* backtracking generated statements:Second is fourth.*/
A kind of algorithm by the extensive clause of learning sample clause acquisition is represented with production system
Learning sample sentence is:
He smoothly answers three problems of foreign guest
=(he, smoothly answers, three problems of foreign guest)
=(he, answer (it answers, IS, fluent), foreign guest's (foreign guest, HAS, three problems))
=(he, answer (it answers, IS, fluent), foreign guest (foreign guest, HAS, problem (problem, IS, three)))
=(he, answers, foreign guest (foreign guest, HAS, problem) ∧ problems (problem, IS, three)) ∧ (it answers, IS, fluent)
=(he answers, foreign guest ∧ problems) ∧ | (foreign guest, HAS, problem) ∧ (it answers, IS, fluent) | ∧ (problem,
IS, three)
Bottom:(problem, IS, three)
The second layer:(foreign guest, HAS, problem) ∧ (it answers, IS, fluent)
Top layer:(he answers, foreign guest) ∧ (he answers, problem)
=(he answers, foreign guest ∧ problems)
Merge:(foreign guest, HAS, problem) ∧ (problem, IS, three) → (he answers, foreign guest ∧ problems)
Note:" answer " determines the characteristic of " fruit " of (he answers, foreign guest ∧ problems) in full sentence causality.
The sample sentence sentence pattern is:
Determiner [x] noun (WHAT1), adverb [y] verb (DO), determiner [z] noun (WHAT2)
Determiner [y]=adverb [y] --- --- [y] be with the matched expression behavior of behavior word to a certain degree,
The determinant generic word of characteristic, state
Determiner [y]=adverb [Y1, Y2 ...]
=(DO, BE, Y1 ∧ Y2 ...)
A kind of algorithm of extensive clause is obtained by learning sample clause:
1) top layer (he answers, foreign guest ∧ problems) is the center clause of sentence.The value of function DO is it is known that i.e. constant " returns
Answer ", the extensive formula of the clause is (x, DO, y).The sentence of other values for containing same functions DO is set as
(S1, S2, S3..., Sj)
Then
(S1, S2, S3..., Sj) in each sentence SjClause (x, DO (answer), y) must be contained.
IF sample sentence center clauses are (x, DO, y)
THEN(S1, S2, S3..., Sj) in search for this clause, obtain (x, y);/ * this be set.*/
X and y is variable.
With IF THEN production system Rule Expression:
IF sample sentence center clauses, THEN (x, DO, y)/* */
IF(S1, S2, S3..., Sj), THEN (x, DO, y)
IF DO, THEN (x, DO, y)/* with DO (answer) be pointer, in sentence collection (S1, S2, S3..., Sj) in search son
Sentence (x, DO (answer), y), obtain (x, y) */
2) second layer:(foreign guest, HAS, problem) ∧ (DO (answer), IS, fluent)
The extensive formula of the clause is (y1, HAS, y2) ∧ (DO, IS, state).
/*Y2∈y1Y=(y1∧y2), (DO, IS, fluent)=(smoothly, DO), " smoothly " is description determiner;
State is the defined domain of function DO, and the value of state integrates element as matter, amount determinant and description ingredient, it may be possible to
Word or phrase.*/
IF second layer clauses are (y1, HAS, y2) ∧ (DO, IS, state)
THEN(S1, S2, S3..., Sj) in search clause y=(y1∧y2) and (state ∧ DO);/ * clause (y1, has, y2)
∧ (DO, IS, state) corresponding forms in sentence S are y=(y1∧y2) and (state ∧ DO).Matter, amount determiner are such as
" good ", " slowly " etc., description word as the * such as " red ", " simple () "/
y1、y2It is variable with DO.
With IF THEN production system Rule Expression:
IF(y1, HAS, y2) ∧ (DO, IS, state), THEN (S1, S2, S3..., Sj) in search clause y=(y1∧y2) and
(state∧DO)/*(y1∧y2) form is (y in natural sentence1y2);(DO, IS, state) be (state DO) */
IF y1y2, THEN (y1, has, y2)
IF(y1, has, y2), THEN (y1∧y2)
IF state DO, THEN (DO, IS, state)
IF (DO, IS, state), THEN (state ∧ DO)
3) bottom:(problem, IS, three)
The extensive formula of the clause is (y2, IS, state).
IF bottom clauses are (problems, IS, three)
THEN(S1, S2, S3..., Sj) in search clause (state ∧ y1);/ * state=(ST1, ST2, ST3...,
STj)*/
Constant " three " is substituted with state, obtains the value set representations of state.
With IF THEN production system Rule Expression:
IF state y2, THEN (y2, IS, state)
IF y2, THEN state
4) successively recall according to sample clause cutting level.
With the corresponding clause of the extensive Shi Ti Change sample clause of each clause:
(he answers, foreign guest ∧ problems) ∧ | (foreign guest, HAS, problem) ∧ (it answers, IS, fluent) | ∧ (problem, IS,
Three)
=(x, DO, y1∧y2)∧|(y1, HAS, y2) ∧ (DO, IS, state) | ∧ (y2, IS, state)
=(he, answers, foreign guest (foreign guest, has, problem) ∧ problems (problem, IS, three)) ∧ (it answers, IS, fluent)
=(x, DO, y1(y1, HAS, y2)∧y2(y2, IS, state)) ∧ (DO, IS, state)
=(he, answer (it answers, IS, fluent), foreign guest (foreign guest, HAS, problem (problem, IS, three)))
=(x, DO (DO, IS, state), y1(y1, HAS, y2(y2, IS, state)))
=(he, smoothly answers, three problems of foreign guest)
=(x, state1DO, y1state2y2)
=he smoothly answers three problems of foreign guest
=x state1DO y1state2y2
With IF THEN production system Rule Expression:
1. IF DO, THEN y1∧y2
IF x, THEN DO
IF x, THEN (DO, y1∧y2)
IF x, THEN DO (DO, IS, state)
2. IF HAS, THEN y2
IF y1, THEN (HAS, y2)
IF(y1, HAS, y2), THEN y1y2
IF y1, THEN (HAS, y2(y2, IS, state))
3. IF DO, THEN state
IF (DO, IS, state), THEN state DO
4. IF x, THEN (DO, y1(y1, HAS, y2)∧y2(y2, IS, state)) ∧ (DO, IS, state)
5. IF x, THEN (DO (DO, IS, state), y1(y1, HAS, y2(y2, IS, state)))
6. IF x, THEN (state1DO, y1state2y2)
7. IF (x, state1DO, y1state2y2), THEN x state1DO y1state2y2
Establish the knowledge base as thinking background automatically by the use of production system
With form be IF the production rule of THEN convert arbitrary natural language sentence.
The natural language got by reading Xi ﹙ directly memories and the Zhi Shi ﹚ got by reasoning by intelligent body thinking Yue
Speech like predicate calculus form sentence, by IF THEN production rule Zhuan Change, just constitute the simple of one rule of knowledge base
Rule.It if desired, can be by it with programming in logic automated programming.
By definition statement WHAT1BE WHAT2, have
IF WHAT1, THEN What2.
What2 is WHAT2 or its variant.
(IFX, THEN Y)
For=(X=Y)/* when substituting calculating, the two is of equal value.*/
=X BE Y
=X Y
Since defined terms " BE " both sides of definition statement WHAT1BE WHAT2 are of equal value, so, the two reciprocal causation:
By definition statement WHAT1BE WHAT2, have
IF WHAT2, THEN What1.
What1 is WHAT1 or its variant.
Equally, in linguistic context instantly,
By WHAT1DO WHAT2, have
IF WHAT1, THEN DO What2.
What2 is WHAT2 or its variant.
By WHAT1DO WHAT2, have
IF DO WHAT2, THEN What1.
What1 is WHAT1 or its variant.
By WHAT BE WHERE, have
IF WHAT, THEN WHERE.
By WHAT BE WHERE, have
IF WHERE, THEN WHAT.
By WHAT BE STATE, have
IF WHAT, THEN STATE.
By WHAT BE STATE, have
IF STATE, THEN WHAT.
The production system Rule Expression of intelligent body learning outcome storage
After intelligent body learns following sentence, it is abstract clause this Yu Ju Zhuan Change and is stored in knowledge base,
Using as learning outcome.The identical correlative of keyword is stored in same storage unit " box ".
Intelligent body learns example sentence:I has criticized her, she cries.
I has criticized her, she cries
=(I, criticizes, she) → (she, cries)
Note:(I, criticizes, she)=(I, it is right ... to criticize, she)
With IF THEN production system Rule Expression:
1. IF (I, criticizes, she), THEN (she, cries)/* (I, criticizes, she) it is to move ahead to state,
(she, cries) is subsequent rear behavior statement.Pay attention to chronological order.*/
This rule of IF (she, cries), THEN (I, criticize, she)/* has such use:If it searches
(she, cries), then can then find (I, criticizes, she).*/
2. IF I, THEN (criticize, she)
IF (criticize, she), THEN I/* this rule has such use:If search is " who " (criticize, she),
Can then find be " I " and " I " other appellations equivalence person.*/
IF is criticized, THEN she
IF is criticized, THEN she/* criticized she=(criticize, she) ∧ (criticizing, BE) */
IF is criticized, THEN
3. IF she, THEN (crying)
IF (crying), THEN she
IF cries, THEN/* " "=completion */
IF is walked, and THEN cries/and * cries=(walking, BE cries) */
IF cries, and THEN is walked
When " crying " this vocabulary is searched in knowledge base, in addition to " IF cries, and THEN is walked " this rule, perhaps may be used also
To collect another rule:" IF cries, and THEN is said ", if " IF cries, and THEN is said " and sentence are bunched You Guan Lian ﹙ i.e. and
" crying, say " mutually closes ﹚, then can draw as the material for generating new sentence.This is related to the rule of selection.
General formula:(determiner [x] WHAT1, determiner [y1] DO1, determiner [z1] WHAT2) →
(determiner [z2] WHAT2, determiner [y2] DO2, determiner [z3] WHAT3)
The IF of this general formula THEN production system regular expression:
1. IF (determiner [x] WHAT1, determiner [y1] DO1, determiner [z1] WHAT2), THEN
(determiner [z2] WHAT2, determiner [y2] DO2, determiner [z3] WHAT3)
IF determiner 2. [x] WHAT1, THEN (determiner [y1] DO1, determiner [z1] WHAT2)
IF determiner [y1] DO1, THEN determiner [z1] WHAT2
IF determiner 3. [z2] WHAT2, THEN (determiner [y2] DO2, determiner [z3] WHAT3)
IF determiner [y2] DO2, THEN determiner [z3] WHAT3
" I has criticized her to sentence, she cries." it is a statement, but can also be used as " knowledge ".As knowledge, this
It is one of reaction for bearing actor caused by the behavior of " criticism ".Its general formula can as " criticism " knowledge it
One is stored in knowledge base " criticism " entry under one's name.And the successive for passing through same words or related term is coupled, and searches different vocabulary
And sentence.
Being abstracted and being specifically defined for " criticism " is one of knowledge about itself.
This is significant to the structure of knowledge base as a principle.
By (determiner [x] WHAT1, determiner [y1] DO1, determiner [z1] WHAT2), equivalence is obtained
Passive type:
(determiner [z1] WHAT2, by determiner [x] WHAT1determiner [y1] DO1)
Or (determiner [z1] WHAT2, by determiner [y1] DO1)
Then have
IF is by determiner [y1] DO1, THEN determiner [z1] WHAT2
By form connection:
IF determiner [z1] WHAT2, THEN ...
Due under linguistic context instantly, IF ... and THEN ... are of equal value, thus IF the production rule quantity of THEN be rendered as idol
Number.When the keyword in the external sentence as pointer or clause, which enter knowledge base, scans for matching, efficiency can be higher.
Example 1:Read statement if (building, have, style) ∧ (style be, succinct) ∧ (building is, modern).This
Three sentences as knowledge storage after former sentence " modern architecture has succinct style " cutting by entering knowledge base.
Modern architecture has succinct style
=(building, have, style) ∧ (style be, succinct) ∧ (building is, modern)
Then there is one group of 6 production rule:
1. IF is built, THEN has style;
IF has style, THEN buildings;
2. IF styles, THEN are succinct;
IF is succinct, THEN styles;
3. IF is built, the THEN modern times.
The IF modern times, THEN buildings.
Example 2:By
Little Yan is accounted for from spring to autumn there are one nest
=(little Yan occupies, a nest), ∧ (occupied, from spring) ∧ (spring arrives, autumn)
Then there is one group of 6 production rule:
1. IF little Yan, THEN are accounted for, there are one nests;
IF is accounted for there are one nest, THEN little Yan;
2. IF occupies, THEN is from spring;
IF occupies from spring, THEN;
3. IF is from spring, THEN to autumn.
IF is to autumn, and THEN is from spring
Situation calculus (situation calculus) is a kind of meaning for the result that state, action and action act on state
The formalization of word calculation.The example of one block world is often used to expression status calculation.If the block world often illustrated
In original state mark be set as S0 (referring to《Artificial intelligence》(Gunnar Nilsson work) p227).Other state marks in Same Scene
Knowledge is set as S1, S2, S3 ....S0 is described like predicate calculus form with natural language below:
(B, on, A) ∧ (A, on, C) ∧ (C, on, Fl) ∧ (B, BE, clear) ∧ (Fl, BE, clear)
Change is into predicate calculus form, i.e.,
On (B, A) ∧ on (A, C) ∧ on (C, Fl) ∧ clear (B) ∧ clear (Fl)
In order to describe this state in situation calculus and other states, we have some abstract concept by state
Body is expressed.
It with formula construction one in state S0 is genuine sentence that here, which is,:
On (B, A, S0) ∧ on (A, C, S0) ∧ on (C, Fl, S0) ∧ clear (B, S0)
We can also be stateful proposition true value.Such as:
With
(Clear (x, s) expression meaning be:X is upper can to put some thing.) with these general axioms it can prove S0's
Various propositions.For example, we can prove thatWith Clear (Fl, S0), etc..
Note:It is to contain symbol.A contains B, is denoted asIt is a proper subclass of A to represent B.
Because of predicate calculus mechanism immediate reasoning state and action, just as in all predicate calculus reasonings, although
Search is still necessary, but search is spatially carried out in a logical expression rather than in a world state now
The model space on carry out.So, natural language can be searched for then like predicate calculus form with immediate reasoning state and action naturally
With expressed by the possibility situation of constitution element permutation and combination.This is a state description space.
Expression action and its result can take following step:
1) behavior act in scene is expressed (i.e. we, which imagine, has such a thing come as action) with sentence.Behavior
Action can be represented with constant, variable or function expression.In situation calculus form, an action is looked at as what action was related to
The function of entity.For example, in the example of common mobile building blocks, action is the function of building blocks.For example, it is contemplated that a building blocks
The action moved from one place to another.
Conventional method is with expression formula move (B, A, Fl)
Represent that the action that B is moved on to from A on floor (it is " right to see an action as one given by that functional value
As ").Usually, pattern (schema) move (x, y, z) can be used to express a move series of acts, wherein x, y and z is mould
Formula variable.Using constant, the instantiation of these variables can generate the expression formula for censuring really action example.(reference《Artificial intelligence
Energy》(Gunnar Nilsson work) p228)
According to this method natural language position is represented like predicate calculus primitive formula (B, move, from A to Fl)
It moves.
(B, move, from A to Fl)
=(B, move, ((B, from, A) ∧ (B, to, Fl)))
A move series of acts is expressed with pattern (schema) (x, move, from y to z), wherein x, y and z is mould
Formula variable.Using constant, the instantiation of these variables can generate the expression formula for censuring really action example.
2) function constant a do or be, it censures a function that action and state can be mapped to state.If a
Represent an action, one state of behalf, then do (a, s) is censured action-state to being mapped to the shape by being censured in s
The function of state obtained by the action that a is censured is performed in state.
This function is expressed as (W (a), do, S (s)) like predicate calculus form with natural language according to this method.With nature
Language is exactly (WHAT, DO, STATE) like the expression of predicate calculus form.
3) result acted with well-formed formula expression.In certain forms, to each action-stream to there are two this box-like
Formula (at present, we ignore action convection current do not have it is influential to).Right to { On, move }, well-formed formula is
It is exactly like the expression of predicate calculus form with natural language:
With
It is exactly like the expression of predicate calculus form with natural language:
We assume here that all variables mentioned in formula are all quantified by full name.
With IF THEN production system Rule Expression:
1. IF s, THEN (x, on, y)
IF x, THEN (on, y)
IF on, THEN y
2. IF s, THEN (x, BE, clear)
IF x, THEN clear
3. IF s, THEN (z, BE, clear)
IF z, THEN clear
④THEN ((x, on, z) ∧ (s, BE, ((x, move, from y to z) ∧ (x, move,
From z to y) ∧ (y, move, from x to z) ∧ (y, move, from z to x) ∧ (z, move, from x to
Y) ∧ (z, move, from y to x)))
Or
5. IF s, THEN ((x, move, from y to z) ∧ (x, move, from z to y) ∧ (y, move, from
X to z) ∧ (y, move, from z to x) ∧ (z, move, from x to y) ∧ (z, move, from y to x))
6. IF x, THEN (move, from y to z)
7. IF x, THEN (move, from z to y)
8. IF y, THEN (move, from x to z)
9. IF y, THEN (move, from z to x)
10. IF z, THEN (move, from x to y)
(11) IF z, THEN (move, from y to x)
Illustrate that transformed sentence is carried out operation principle and the work of creative thinking study by unit in detail below
Process:
Production system represents the connection logic for the concept that similarity relation causes.
For example, fragmentary fine granularity jade is similar with the dewdrop in early morning, then it can " broken jade " replacement " dew ".
Target conclusion:IF (fragmentary fine granularity jade and ... similar, the dewdrop in early morning), THEN (" broken jade " substitutes,
" dew ")
Reasoning process:
Fine granularity jade fragmentary IF, THEN (and ... similar, the dewdrop in early morning)
IF (dewdrop in early morning and ... similar), fine granularity jade fragmentary THEN
Fine granularity jade fragmentary IF, THEN and ... it is similar
The dewdrop in IF early mornings, THEN and ... it is similar
The dewdrop in IF early mornings, fine granularity jade fragmentary THEN
IF and ... it is similar, THEN seemingly ...
The dewdrop in IF early mornings, THEN (seemingly, fragmentary fine granularity jade)
The dewdrop in IF early mornings, THEN is like fragmentary fine granularity jade
Obtain sentence:The dewdrop in early morning is like fragmentary fine granularity jade.
The dewdrop in IF early mornings, the dewdrop=dew in THEN dew/* early mornings, the two are of equal value.*/
Fine granularity jade fragmentary IF, fine granularity the jade fragmentary broken jade/* of THEN=broken jade, the two are of equal value.*/
IF (dewdrop in IF early mornings, fine granularity jade fragmentary THEN), THEN (IF reveals, the broken jade of THEN)
IF reveals, the broken jade of THEN
Revealed=broken jade, the two is of equal value.
Game like predicate calculus form represent Zhuan Change be IF THEN production system Rule Expression
Two people a, b play stone, scissors, cloth game.
The combination of actions of two sides generates an effectiveness pair.The search that everything combination constitutes game possibility is empty
Between.Also effectiveness (income) matrix is formed.Expressed with the production rule of IF THEN, can direct self-programing, save
Chart.
1) according to constitution element, all permutation and combination are made.
Combine (search) spatial utility (income) matrix (a, b value)
(a stones, right, b stones) ∧ (b stones, right, a stones);=0,0
IF (a stones, right, b stones), THEN 0
IF (b stones, right, a stones), THEN 0
IF a stones, THEN b stones
IF b stones, THEN a stones
IF a value of utilities, THEN 0
IF b value of utilities, THEN 0
(a stones, right, b scissors) ∧ (b scissors, right, a stones);=1, ﹣ 1
IF (a stones, right, b scissors), THEN 1
IF (b scissors, right, a stones), THEN ﹣ 1
IF a stones, THEN b scissors
IF b scissors, THEN a stones
IF a value of utilities, THEN 1
IF b value of utilities, THEN ﹣ 1
(a stones, right, b cloth) ∧ (b cloth, right, a stones);=﹣ 1,1
IF (a stones, right, b cloth), THEN ﹣ 1
IF (b cloth, right, a stones), THEN 1
IF a stones, THEN b cloth
IF b cloth, THEN a stones
IF a value of utilities, THEN ﹣ 1
IF b value of utilities, THEN 1
It is distributed there are two identical thing x1 and x2 by two people a, b, point-score has 24 kinds of combinations (1 × 2 × 3 × 4).
Everyone has 3 kinds of distribution situations:Retain, share, give.
Combine (search) spatial utility (income) matrix (a, b value)
(a retains, x1 and x2) → (b gives, x1 and x2);2,0
IF (a retains, x1 and x2), THEN (b gives, x1 and x2)
IF (b gives, x1 and x2), THEN (a retains, x1 and x2)
IF (a retains, x1 and x2), THEN 2
IF 2, THEN (a retains, x1 and x2)
IF (b gives, x1 and x2), THEN 0
IF 0, THEN (b gives, x1 and x2)
(a shares, x1) → (b shares, x2);1,1
IF the production system regular expression of THEN be same as above.
General formula:(a, DO, x1 and x2) → (X1 and x2) 2,0
(a, DO, x1) → (b, DO, x2) 1,1
Examples below 3 is then to enter knowledge base searching relevant knowledge:
Example 3:External world's input Guan Jian Ci ﹙ Zhi Zhen ﹚ or most simple clauses, search for matched knowledge entry in knowledge base.
If knowledge base has a knowledge:If rained, street surface is wet.
Now, " raining " is inputted, searches for " if raining " in knowledge base, what kind of result had
Existing this knowledge, searches for and matches in knowledge base:
IF rained, and THEN street surfaces are wet
=IF (raining) ∧ THEN (street surface is wet)
=IF (..., lower mistake ..., rain) ∧ THEN (into street surface, being, wet)
=outside has (..., lower mistake ..., rain) or enters knowledge base searching matching (..., lower mistake ..., rain) with valency person
With success → output (street surface is, wet)/* (..., lower mistake ..., rain) and (rain, under, cross) same to valency.If (..., under
Cross ..., rain) or with valency person enter knowledge base searching matching, if successful match, then output one knowledge:(street surface,
It is, wet), i.e., street surface is wet.*/
In turn, production rule transformation easily can be generated as natural language sentence by we.By
IF (..., lower mistake ..., rain) ∧ THEN (street surface is, wet)
=IF (raining) ∧ THEN (street surface is wet)
=﹙ such as fruit ﹚ rained, then that ﹚ street surface of ﹙ is wet
=rain, street surface or wet
=just rained, street surface or wet/* needs build vertical 3 production rules of a Group in knowledge base:
1. IF (..., lower mistake ..., rain) ∧ THEN (street surface, still, wet)
IF 2. (street surface, still, wet) ∧ THEN (..., just descended, rain)
3. IF (..., just descended, rain) ∧ THEN (street surface, still, wet)
3 production rules of this Group, by natural sentence, " if rained, street surface is wet." conversion and
Come.
It restores 3., IF (..., just descended, rain) ∧ THEN (street surface, still, wet)=just rained, street surface
It is or wet
Output.
By restoring 3., equivalent statement is exported, is equivalent to generated statement output.*/
……
We need vocabulary, the statement law of all equivalences to be established in the form of production rule in dictionary or knowledge
In library.
Output unit control intelligent body is finally discussed in detail and makes corresponding actions;Or generated statement output is as conclusion;Or
The operation principle and the course of work stored as learning outcome.
Realize mode of the natural language word sentence to the control of intelligent body
Input by sentence is equivalent to sensor input and (is expressed as s in the example of two-dimensional grid space1..., s88 grid structures
Into environment space feature vector expression) function.It sends out natural language literal order and then generates corresponding sensor input, then
Action behavior after generating intelligent body aware processes is inputted by sensor.In turn, thus sensor input generates that this is corresponding
Natural language word sentence.It is thusly-formed control of the natural language word sentence to intelligent body.It represents as follows:
Intelligent body x northwards advance → advance north →→north
→ intelligent body x northwards advances
Intelligent body x northwards advances
=(intelligent body x northwards, advances)
=(intelligent body x (intelligent body x, to north), advance)
=(intelligent body x advances) ∧ (intelligent body x, to north)
①Then intelligent bodies x northwards advances
If intelligent bodies x northwards advances, then (intelligent body x northwards advances)
If intelligent bodies x, then northwards advance
If intelligent bodies x, then row ∧ into/* (If intelligent bodies x, then row ∧ into)=(intelligent body x, row ∧ into) */
If rows ∧ into, then (to north)/* (If rows ∧ into then (to north))=(row ∧ is into north) */
If to, then north/* (If to then north)=(to north) */
If rows ∧ into, then north/* (If rows ∧ into then north)=north * that advances/
2. If (If rows ∧ is into then is northern),
then north
Intelligent body x westwards advance → advance west →→west
→ intelligent body x westwards advances
……
Etc..
Here is the program controlled with English sentence corresponding with this sentence:
With IF THEN production system regular expression, make intelligent body thinking automated programming, also allow for intelligent body thinking meter
It calculates.
A recurrence semantic structure is in thinking, according to some condition it is true with it is false from two possible situations selection-
It is a.Such example has:
If certain company's production and sales is increased larger.The said firm's stock is so bought, otherwise, sells the said firm's stock
Ticket;
Buy or sell larger growth or larger reduction of certain corporate share depending on the said firm's production and sales
Each such sentence can be write as the form met such as lower structure:
If (condition) then (situation)
Else (situation)
Here, we used keyword if (if), ((otherwise > guides entire main structure to then by so > and else
In different minor structures, and limit using bracket the boundary of minor structure.Using this syntactic structure as pseudocode, we
The unification side for allowing intelligent body thinking using natural language like the sentence structure of predicate calculus form expression human thinking is just obtained
Method.Sentence below
Whether it is the leap year according to the time, total number of days is correspondingly removed by 366 or 365.
If (time is the leap year), then (total number of days ← total number of days is removed by 366),
Else (total number of days ← total number of days is removed by 365)
The If times, the then leap years,
If leap years, the total number of days of then
The total number of days of If, then (total number of days, quilt ... remove, 366)
The total number of days of If, then (quilt ... removes, 366)
If quilts ... remove, then366
Else (total number of days ← total number of days is removed by 365)
If else times, then (total number of days ← total number of days is removed by 365)
The total number of days of If, then (total number of days, quilt ... remove, 365)
The total number of days of If, then (quilt ... removes, 365)
If quilts ... remove, then365
We can also use shorter grammer:
If (condition) then (situation)
This form is not related to situation otherwise.Utilize this representation, sentence
If in sales volume in the regional markets of reduction, then selling price lowers 5%.
It can be reduced to.
If (regional markets sales volume is being reduced), then (selling price is lowered, 5%)
If (regional markets sales volume), then are reduced
If is reduced, then (selling price is lowered, 5%)
If selling prices, then (are lowered, 5%)
If is lowered, then 5%
For sentence of the natural language like predicate calculus form, this form is with the most use.
The pseudocode of a sentence pattern represents from the point of view of us:
1. has following sentence pattern
WBW patterns (" WHAT BE WHAT " pattern)
=(determiner [x] noun (WHAT1), BE, determiner [z] noun (WHAT2))
2.IF noun (WHAT1), BE, noun (WHAT2) are it is known that THEN (WHAT1) is of equal value with (WHAT2);
IF (WHAT1) and (WHAT2) equivalence THEN (IF (WHAT1) THEN (WHAT2)).
IF noun (WHAT1), BE, noun (WHAT2) are with being it is known that THEN [x], [z] is limits composition.
AND IF [x], [z] are set of words, and THEN [x], [z] limits word set for table-like.
Determin Table-like qualifier to[x]and[z]when the previous WHAT1and
WHAT2is NN
Else mark masters have pronoun (possessive pronoun)./ * such as " my noun (WHAT1) " */
3. returning to 1, marked for new input.
NN:Noun,sing.or mass.such as llama
Limiting composition in the basic thoughtcast of natural language clause, in sentence can regard what is determined by being defined noun as
The intersection of scope and the scope by being determined as the adjective of attributive function.Be the equal of cheap east in example so below
The intersection of the scope in west and the scope in restaurant.
The restriction composition of determiner [x] noun (WHAT1) has corresponding form, i.e.,
IF determiner [x] noun (WHAT1), THEN
determiner[x]noun(WHAT1)
=(noun (WHAT1), BE, determiner [x]) ∨ (noun [x], HAS, noun (WHAT1))
THEN (noun (WHAT1), BE, determiner [x])
=noun (WHAT1BE determiner [x]
Here is example:(quoted from《The comprehensive opinion of natural language processing》(U.S.) Jurafsky.D, p355)
1don't mind a cheap restaurant.
This restaurant is cheap.
A cheap restaurant=This restaurant is cheap
Equally, the restriction composition of determiner [y] verb (DO) has corresponding form, i.e.,
IF determiner [y] verb (DO), THEN
determiner[y]verb(DO)
=(noun (DO), BE, determiner [y])/* noun (DO) represent variables D O be now noun */
Note:
What former was expressed in former friend (pervious bosom friend, old to know) is to limit the time.
Former friend=(friend, BE, former), No is problematic in the sense for this statement.
The new sentence of natural language and the production system of phrase generation represent
The production system Rule Expression that the meaning of a word of word is explained in dictionary, example sentence and its new sentence and phrase generate
[cigarette]={ balance spring } floats balance spring in the air, the cigarette of balance spring, that is, fine.
﹙ is aerial, floats, You Si ﹚ ∧ ﹙ balance springs, that is, fine Yan ﹚
=﹙ are aerial, float, You Si ﹙ balance springs, that is, fine Yan ﹚ ﹚
=﹙ are aerial, float, fine Yan ﹚/* it is synonymous sentence element --- for example phrase is substituted for each other.*/
=float fine cigarette in the air
See opposite dicing process, the pointer for paying attention to being coupled two clauses is " balance spring " word:
The aerial fine cigarette that floats
=﹙ are aerial, float, fine Yan ﹚
=﹙ are aerial, float, You Si ﹙ balance springs, that is, sentence element --- for example phrase phase synonymous fine Yan ﹚ ﹚/*
Trans-substitution.*/
=﹙ are aerial, float, You Si ﹚ ∧ ﹙ balance springs, that is, fine Yan ﹚
General formula:what1DO what2∧what2BE what3
=what1DO what3
IF what1DO what2 ∧ what2BE what3, THEN what1DO what3
IF what1, THEN DO what2
IF DO, THEN what2
IF what2, THEN BE what3
IF what2, THEN what3
IF ﹙ balance springs, that is, Yan ﹚ of emerging in the chimney on the lonely room on wilderness that towers from above all others,
THEN ﹙ are aerial, float, Yan ﹚ of emerging in the chimney on the lonely room on wilderness that towers from above all others
IF ﹙ are aerial, float, Yan ﹚ of emerging in the chimney on the lonely room on wilderness that towers from above all others,
The cigarette emerged in the chimney that THEN is floated in the air on the lonely room to tower from above all others on wilderness
The Yu Ju Zhuan Change of metaphor expression are production system Rule Expression
Example sentence:She has clean body as one from the thing having not been used.
She has clean body as one from the thing having not been used
She has a clean body to=IF, and THEN is as one from the thing having not been used
=IF (she, has, body) ∧ (body, IS, ∧ clean) ∧ (she, IS is clean)), THEN (she, as thing
(thing, IS, a ∧ from have not been used))
=IF (she, has, body) ∧ (body, IS, ∧ clean) ∧ (she, IS is clean)), THEN (she, as east
West) ∧ (thing, IS, a ∧ from have not been used)
By " she has a clean body ", metaphor sentence is drawn:" just as one from the thing having not been used ".
Original state:She has a clean body
Dbjective state:She is as one from the thing having not been used.
She has a clean body → she is as one from the thing having not been used
=(she, has, a clean body) → she is as one from the thing having not been used
=(she, has, (body, IS, ∧ clean)) → she is as one from the thing having not been used
=(she, has, body) ∧ (body, IS, ∧ clean) ∧ (she, IS is clean)) → she as one never by
Used thing.
Wherein (body, IS, ∧ clean) ∧ (she, IS is clean) with 3. in premise conclusion " x is clear " phase
Together, then by
(she, has, body)=(X has, thing) --- --- sees premise sentence above 3.,
Therefore premise sentence 3. can be drawn, and " she as " is added before sentence in the database, here it is sentences
" she is as one from the thing having not been used.”
Note:
" body " is object in (she, has, body), and " body " belongs to an element or one in set [thing].And " she " is
Synonym, " she " belong to an element or one in set [X].That is
(she, has, body)=(X has, thing) --- --- sees premise sentence above 3.,
This, which can become, determines and 3. one of unique corresponding mark.
With IF THEN production system Rule Expression:
Existing 3 sentences or sub- sentence ﹙ pieces words and phrases group ﹚:
She has a clean body;
She is as thing;
Thing is one and from having not been used.
Following sentence can be formed with this 3 sentences or clause to bunch:
((she, has, body) ∧ (body, IS, ∧ clean) ∧ (she, IS, clean)), (she, as thing) ∧ (east
West, IS, a ∧ are from having not been used)
By " she has a clean body ", metaphor sentence is drawn:" just as one from the thing having not been used ".
1. IF what1, THEN (have, what2)
What2 clean mono- ∧ of IF what2, THEN
What1 clean mono- ∧ of IF what1, THEN
2. IF is 1., THEN is 3.
3. IF what1, THEN (as what)
IF what1, THEN pictures
IF pictures, THEN what
Mono- ∧ of IF what, THEN is from the what having not been used
IF what1, THEN (as a, ∧ is from the what having not been used)
Backtracking generation nature sentence:What1 is as one from the what. having not been used
【Sentence example】I has been arrived outside oneself room, my mother in face of out, has then flown out eight years old already
The macro youngster of nephew.(Lu xun《Native place》)
With IF/THEN production systems Rule Expression, this sentence is bunched;And pass through logical operation and amplify to form new language
Sentence:
I has arrived outside oneself room IF, THEN (my mother already in face of out, the nephew then to have flown out eight years old
Macro youngster);
IF my mothers are already in face of out, the macro youngster of nephew that THEN has then flown out eight years old;
I has been arrived outside oneself room
=(I, arrives, outside oneself room)
=(I, is to (arriving, IS), outside room (outside room, IS, oneself))
=(I, arrives, outside room) ∧ (arriving, IS) ∧ (outside room, IS, oneself)
=(I, arrives, outside room) ∧ (arriving, IS) ∧ (outside room, belong to, oneself)
=(I, arrives, outside room) ∧ (arriving, IS) ∧ (oneself, HAS, outside room)
=(I, arrives, outside (room, HAS, outside)) ∧ (arriving, IS) ∧ (family (family, IS, oneself), HAS, outside
(room, HAS, outside))
=(I, arrives, outside) ∧ (arriving, IS) ∧ (room, HAS, outside) ∧ (family, HAS, outside) ∧ (family, IS, oneself
)
With this sentence of IF/THEN production systems Rule Expression:
IF I, THEN (is arrived, outside room);
IF is arrived, THEN;
IF is arrived, outside THEN rooms;
Outside IF rooms, THEN (belong to, oneself);
IF belongs to, THEN oneself;
Outside IF rooms, THEN oneself;
I=is to outside room
Outside room=oneself/* (outside IF rooms, THEN oneself)=(outside room=oneself), the two is of equal value.*/
It obtains:I arrived oneself/* backtracking generated statement.*/I am to outside room;
By
To=then
I am outside room.
My mother is already in face of out
=(my mother ︱ determiner [x] what1, already in face of out ︱ adv [y] DO, ︱ determiner [x]
What2 the expression of the general formulas of)/*." " is the expression of " behavior completion ".*/
=(my mother, already in face of out)
=(my mother, out (out, IS, already ∧ in face of))
=(my mother, out) ∧ (out, IS, already ∧ in face of)
=(my mother comes out (out, IS, already) in face of ∧)
With this sentence of IF/THEN production systems Rule Expression:
My mothers of IF, THEN (being come out in face of ∧);
My mothers of IF, THEN (my mother, comes out in face of ∧);
My mothers of IF, THEN in face of;
My mothers of IF, THEN (out);
IF comes out in face of ∧, THEN;
IF comes out in face of ∧, and THEN is already;
IF is in face of THEN;
IF comes out, THEN;
It obtains:My mother is already in face of/* backtracking generated statements.*/
My mother has come out/* backtracking generated statements already.*/
My mother in face of I/* backtracking generated statement.*/
My mother in face of to outside room/* I=to outside room, recall generated statement.*/
My mother come out to outside room/* by (IF in face of THEN) and (IF comes out, THEN), in face of=out, two
Person is of equal value, alternative;Recall generated statement.*/
By
My mother come out to outside room ∧ I outside room
=(my mother, out ∧ arrive, outside room) ∧ (I, outside room)
=(my mother, out, outside room) ∧ (my mother, arrives, outside room) ∧ (I, outside room)
=(my mother goes out ∧, outside room) ∧ (my mother, outside room) ∧ (I, outside room)
=(my mother goes out ∧, outside room) ∧ (my mother ∧ I, outside room)
=(my mother goes out ∧, outside room) ∧ (my mother ∧ I, outside room)
(my mother goes out ∧, outside room)
=(my mother, goes out, outside room) ∧ (my mother comes, outside room)
=my mother goes out my mothers of ∧ outside room and comes outside room
(my mother ∧ I, outside room)=my mother and I outside room
By (my mother ∧ I, outside room), then
IF my mother ∧ I, THEN (outside room)
IF my mother ∧ I, THEN exists
IF exists, outside THEN rooms
IF my mother ∧ I, outside THEN rooms
A general rules of IFx ∧ y, THEN (in the Where)/* from knowledge base.*/
The special case of a general rules of IF Where, THEN (the together)/* from knowledge base.*/
IFx ∧ y, THEN are together
﹙ x ∧ y are outside room | (Where) → x ∧ y together, outside room | (Where) expression be " outside room " set (Where) in
An element.﹚
So have
IF my mother ∧ I, THEN is together
My mother with I together with.
Then the macro youngster of nephew to have flown out eight years old
=(then, just flying out, the macro youngster of nephew of eight years old)
=(then ..., just fly out, the macro youngster of nephew (the macro youngster of nephew, IS, eight years old))
=(then (my mother, out), fly, the macro youngster) ∧ ﹙ of nephew fly, IS, just ... Chu ﹚ ∧ (the macro youngster of nephew,
IS, eight years old)
With this sentence of IF/THEN production systems Rule Expression:
Then (my mother, out), THEN (flies, the macro youngster) ∧ ﹙ of nephew fly, and IS, just ... (nephew is macro by Chu ﹚ ∧ IF
Youngster, IS, eight years old);
Then (my mother, out), THEN flies IF
IF flies, the macro youngster of THEN nephews;
IF flies, THEN just ... go out;
The macro youngster of IF nephews, THEN eight years old;
IF flies, the macro youngster ∧ of THEN nephews just ... go out;
IF goes out, THEN out/both * is synonymous;*/
IF flies, the macro youngster ∧ of THEN nephews just ... out;
IF then (my mother, out), THEN (the macro youngster of nephew, just comes out)
By
Winged=macro the youngster of nephew/* IF fly, and the macro youngster * of THEN nephews/
Fly=just ... gone out/* IF fly, THEN just ... gone out */
It obtains:The macro youngster of nephew just gone out/* IF fly, the macro youngster ∧ of THEN nephews just ... gone out */
It obtains:Then the macro youngster of nephew just comes out
=then ... ∧ (the macro youngster of nephew, just comes out)/* IF go out, THEN comes out, and the two is synonymous.For syllable word of making sentences
Custom, with " coming out " replacement " going out ".*/
=then ... ∧ (the macro youngster of nephew, out (out, IS, just))
=then ... ∧ (the macro youngster of nephew, out) ∧ (out, IS, just)
IF then (my mother, out), THEN (the macro youngster of nephew, out) ∧ (out, IS, just)
IF (my mother, out), THEN (the macro youngster of nephew, out)
It obtains
IF (my mother, out) ∧ (the macro youngster of nephew, out), (my macro youngster of mother ∧ nephews, goes out THEN
Come),
My macro youngsters of mother ∧ nephews of IF, THEN together, are obtained
Together with my mother and the macro youngster of nephew;
" my mother with me together with " has been generated above, has been obtained after merging:
My macro youngster of mother, nephew with I together with.
【Reasoning example】If I am at that time above deck, if I passes by here just, perhaps I can see her.
If=(I, existed at that time ... on, deck) if ∧ (I, passes by just, here) → (perhaps I, can see,
She)
If=((I, existed at that time ... on, deck) ∧ (I, passes by just, here)) → (meeting) (perhaps I, see,
She)
If=((I, existed at that time ... on, deck) ∧ (I, passes by just, here)) → (meeting) (I, sees, she)
∧ (sees, is, perhaps)
IF THEN production system regular expression:
1. IF (I, existed at that time ... on, deck) ∧ (I, passes by just, here), THEN (meeting) (I, sees, she) ∧
(see, be, perhaps)/* (meeting) is the hypothesis statement to whole sentence, (she, cries) it is subsequent rear behavior statement.Pay attention to
Chronological order overturns --- and by virtually assuming linguistic context, setting now is gone over.*/
IF (meeting) (I, sees, she) ∧ (see, be, perhaps), THEN (I, existed at that time ... on, deck) ∧ (I, just
Pass by well, here)/* */
2. IF I, THEN (on existing at that time ..., deck) ∧ (passing by just, here)
On IF existed at that time ..., THEN decks
IF is passed by just, and THEN is here
IF is passed by, and THEN is lucky
3. IF I, THEN (can see, she)
IF sees that THEN is perhaps
IF sees, THEN she
The sentence of above-mentioned statement, if with IF the production system regular expression of THEN its general formula, it is possible to as " false
If one of knowledge of statement " is stored in knowledge base " assuming that statement ", entry is under one's name.
(if I existed ...=(on existing at that time ..., deck)=at that time goes up=deck) ∧
(my=(passing by just, here (deck))=pass by just=here (deck)) →
(meeting) (I=(it sees, she)=see ≠ she) ∧ (see=perhaps)
" she " only the behavior with her and result or certain restriction are of equal value.But " she " has transitivity, i.e. " I " and " she "
It is substantially indirectly of equal value.
﹝ Shi Li ﹞
" Sam (name of an agent) knows building blocks A above B ", we write:
Sam (name of an agent) knows building blocks A above B
=(Sam, it is known that, (building blocks A, on, building blocks B))
=(Sam, it is known that) ∧ (know, BE, (building blocks A, on, building blocks B))
X [Agent1, X (Agent2, On (A, B))], it means that Agent1 knows that Agent2 knows A on B.
" Mike (name of an agent) knows that Sam (name of an agent) knows building blocks A above B "
=(Mike, it is known that2, ((Sam, it is known that1) ∧ (knows1, BE, (building blocks A, on, building blocks B))))
=(Mike, it is known that2) ∧ (knows2, BE, ((Sam, it is known that1) ∧ (knows1, BE, (building blocks A, on, building blocks B))))
=(Mike, it is known that2) ∧ (knows2=((Sam, it is known that1) ∧ (knows1=(building blocks A, on, building blocks B))))
Zhuan Change are represented for production system:
If (Mike, it is known that2), then (knows2=((Sam, it is known that1) ∧ (knows1=(building blocks A, on, building blocks B)))
If Mike, then know2
If knows2, then ((Sam, it is known that1) ∧ (knows1=(building blocks A, on, building blocks B)))
If knows2, then (Sam, it is known that1)
If knows2, then Sam
If knows2, then knows1
If knows2, then (knows1=(building blocks A, on, building blocks B))
If knows1, then (building blocks A, on, building blocks B)
If (Sam, it is known that1), then (knows1=(building blocks A, on, building blocks B))
If Sam, then know1
If knows1, then (building blocks A, on, building blocks B)
If knows2, then (building blocks A, on, building blocks B)
(If knows2, then (building blocks A, on, building blocks B))=(know2, BE, (building blocks A, on, building blocks B)), substitute into (Mike,
Know2), it obtains
(Mike, it is known that2)
=(Mike, it is known that2(know2, BE, (building blocks A, on, building blocks B)))
=(Mike, it is known that2, (building blocks A, on, building blocks B))
=Mike knows building blocks A on building blocks B
【Plan example】Original state and target
Original state by it is specified initial when each feature value define.
Forward direction is planned
It is represented with production system of the natural language like predicate calculus form, it can be empty to avoid the search that conventional method is brought
Between complexity.Enumerate see former book example 8-8 (《Artificial intelligence:Calculate Agent bases》, the works such as David L.Poole, p234)
The original state original book of search space represents:
(intelligent body people Rob, in cafe)=cs
∧ (Sam, it is desirable to, coffee)=swc
∧ (office for incoming and outgoing mail has, mail)=mw
The natural language of the original state (former book) of the search space is represented like the production system of predicate calculus form:
If intelligent body people Rob are in cafe, then (intelligent body people Rob, in cafe)
If intelligent body people Rob, then (in cafe)
If (in cafe), then intelligent body people Rob/* (If (in cafe), then intelligent body people Rob)=in coffee
The intelligent body people Rob in coffee shop, you can generate this new sentence phrase.*/
If exists, then caves
If caves, then/* (If caves, then exist)=cafe whereThis newspeak can be generated
Sentence.*/
If intelligent body people Rob, then caves
If caves, the intelligence in then intelligent body people Rob/* (If caves, then intelligent body people Rob)=cafe
Body people Rob, you can generate this new sentence phrase.*/
If intelligent body people Rob No have coffee, then (intelligent body people Rob , No have, coffee)
If intelligent body people Rob, then (No have, coffee)
If (No have, coffee), then intelligent body people Rob/* (If (No have, coffee), then intelligent body people Rob)=No have
The intelligent body people Rob of coffee, you can generate this new sentence phrase.*/
If No have, and then coffees/* (If No have, then coffees)=No has coffee, you can generates this new sentence.*/
If coffees, then No have/* (If coffees, then No have)=(coffeeNo has), you can generate this new sentence piece
Language.*/
If Sam want coffee, then (Sam, it is desirable to, coffee)
If Sam, then (want, coffee)
If (wants, coffee), the Sam of then Sam/* (If (wants, coffee), then Sam)=desired coffee, you can
Generate this new sentence phrase.*/
If wants, then coffees
If coffees, then wants/and * (If coffees, then want)=coffee wants ∨ coffees and is intended to ∨ coffees to be intended to
, you can generate this new sentence.*/
There are mail, then in If office for incoming and outgoing mails (office for incoming and outgoing mail has, mail)
If office for incoming and outgoing mails, then (have, mail)
If (has, mail), and then office for incoming and outgoing mails/* (If (has, mail), then office for incoming and outgoing mails)=there is mail in office for incoming and outgoing mail, you can
Generate this new sentence.*/
If has, then mails/* (If has, then mails)=have mailThis new sentence can be generated.*/
If mails, then has/and * (If mails, then have)=mail has ∨ mails to haveThis newspeak can be generated
Sentence.*/
If intelligent body people Rob No have mail, then (intelligent body people Rob , No have, mail)
If intelligent body people Rob, then (No have, mail)
If (No have, mail), then intelligent body people Rob/* (If (No have, mail), then intelligent body people Rob)=No have
The intelligent body people Rob of mail, you can generate this new sentence phrase.*/
If No have, and then mails/* (If No have, then mails)=No have mailThis new sentence can be generated.*/
If mails, then No have/and * (If mails, then have)=You Jian No have ∨ You Jian No to haveIt can generate this new
Sentence.*/
The natural language of the intermediate state of the search space is represented like the production system of predicate calculus form:
1. If Sam No have coffee, then Sam want coffee
=If (Sam , No have, coffee), then (Sam, it is desirable to, coffee)
If Sam want coffee, and then Sam want Rob to buy coffee
=If (Sam, it is desirable to, coffee), then (Sam, (Rob is bought, coffee))
=If (Sam, it is desirable to, coffee), then (Sam, Rob) ∧ (Rob is bought, coffee)
=If (Sam, it is desirable to, coffee), then (Sam, Rob) ∧ If (Sam, it is desirable to, coffee), then (Rob is bought,
Coffee)
If (Sam, it is desirable to, coffee), then (Rob is bought, coffee)
If (Rob is bought, coffee), then (Sam, it is desirable to, coffee)
2. If (intelligent body people Rob , No have, coffee), then (Rob is bought, coffee)
If (Rob is bought, coffee), then (Rob, in cafe)
If (Rob, in cafe), then (Rob is bought, coffee)
If Rob, then (in cafe)
If Rob, then exist
If Rob, then caves
If exists, then caves
If Rob, then (are bought, coffee)
If Rob, then are bought
If Rob, then coffees
If is bought, then coffees
Note:If originally (intelligent body people Rob, in laboratory), then need to be moved to behavior generation position.Arrive first coffee
Shop or office for incoming and outgoing mail, are related to shortest route problem.
3. If (Sam, it is desirable to, coffee), then (intelligent body people Rob , No have, coffee) ∨ (Rob is bought, coffee)
If (Rob is bought, coffee), then (Rob, coffee) ∧ (coffee ∨ Rob, be sent to, Sam)
=If (Rob is bought, coffee), then (Rob, coffee)
∧ If (Rob is bought, coffee), then (coffee ∨ Rob, be sent to, Sam)
If (Rob, coffee) ∧ (coffee ∨ Rob, be sent to, Sam), then (Rob is bought, coffee)
If (Sam, in the office of Sam), then (coffee ∨ Rob, be sent to, Sam)
If (coffee ∨ Rob, be sent to, Sam), then (Sam, in the office of Sam)
=If (coffee ∨ Rob, be sent to, Sam), then (coffee ∨ Rob, are sent to, the office of Sam)/* Sam=Sam's
Office, the two equivalence relation.
If Sam, then (in the office of Sam)
The Ban Gong Shi ﹚ meanings of office , ﹙ If Sam, the then Sam of If Sam, then Sam are the offices of Sam=Sam
Room.The two is of equal value.*/
" office of Sam=Sam " substitutes into above formula, has
If (Rob is bought, coffee), then (Rob, coffee) ∧ (coffee ∨ Rob, are sent to, the office of Sam)
4. If (office for incoming and outgoing mail has, mail), then (Rob takes, mail)
If (Rob takes, mail), then (Rob, in office for incoming and outgoing mail)
If (Rob takes, mail), then (Rob has, mail)
If (Sam, mail), then (Rob takes, mail)
If (Sam, mail), then (Rob, mail) ∧ (mail ∨ Rob, be sent to, Sam)
If (Rob takes, mail), then (Rob, mail) ∧ (mail ∨ Rob, are sent to, the office of Sam)
The natural language of the dbjective state of the search space is represented like the production system of predicate calculus form:
5. by 3. and 4., merge two formulas:
If (Rob is bought, coffee) ∧ (Rob takes, mail), then (Rob, coffee ∧ mails) ∧ (coffee ∧ mails ∨
Rob is sent to, the office of Sam)
It can recall and generate new sentence:
(Rob, coffee ∧ mails) ∧ (coffee ∧ mail ∨ Rob, be sent to, the office of Sam)
=(Rob, coffee mail (coffee ∧ mails, are sent to, the office of Sam))
=(Rob sends coffee mail to, the office of Sam)
=(Rob sends coffee mail to, the office of Sam)
=Rob sends coffee mail to the office of Sam
More action details represent to omit.
Represented with production system of the natural language like predicate calculus form, indicating originally is extra, have it is new because
Fruit relative clause then requires supplementation with.
【Solve wisdom problem-instance】The superposition that space limits
One length and another length are that the space of two kinds of difference limits, and the two has the effectiveness or potential folded of superposition
The effectiveness added, the effectiveness after superposition are similar to the length effectiveness for the sum of said two devices.Such as a wooden stick with it is another
There are the effectiveness to link together or potential effectiveness between root wooden stick, when we define the longer body of wooden stick class, at it
The effectiveness dimension of this interconnection should be just included in Gestalt structural dimension.For another example a rope type objects are suspended in some table
Face, Gestalt structural dimension just include:
1) with the vertical relation on ground;
2) anchor point is located at the space restriction scale of upper end and anchor point (with respect to Mr. Yu's object of reference);
3) since it has flexible and lower end only there are one the constraint of space limitations dimension, then its potential physics with pendulum
Function (effectiveness) and in this case its space limited area (amplitude range of pendulum);
4) it suspends the effectiveness of other objects in midair;
5) its own gravity force-bearing situation.
For another example any one object all with or potentially with carry its object Gestalt structural dimension.In above-mentioned rope
In the Gestalt structural dimension of class suspension, 3) in the physics effectiveness of " pendulum " can in addition segment its Gestalt structural dimension again,
Belong to the form tower structure of another things.Wherein it is there are one dimension:Increase by one in one end towards ground of rope type objects
A part weight (object) can enhance the intensity (thus increasing the amplitude and efficiency swung) of " swing ".In a classical psychology
It learns in experiment " two strings problems " and is related to this principle.[8]
By above-mentioned, we can establish the frame of a thought process, on this basis, in each part of frame
(part) forms the subsystem of a drawer type, and the set of patterns of some self-organizing system structure is put into wherein, this set of patterns
Many of relevant different mode.This is realized by below step:
Establish extensive expression formula;
With most simple thoughtcast statement thought process;
Result of thinking, output are expressed using the sentence set generated statement of most cylinder thoughtcast.
According to this thinking, we " two strings problems " as an example:
1) primary condition:
A, people (its Gestalt structural dimension is various, can self is displacement, etc. wherein having);
People:| (people, can be with, activity) → (people, can be with mobile) ∧ (people, can to grab) | ∧ (people is, independent) ∧
(people is heavy big object) ∧ ...
B, Two roots Sheng Fen Do are located at ceiling, and (its Gestalt structural dimension is mutually the same, wherein consolidating for one end for suspended state
Fixed, the constrained condition of the other end can be changed, and be currently by gravity constraint, perpendicular to ground.Another Gestalt structural dimension is
One end can be done limited without fixation under longitudinal gravity and the effect of horizontal direction power below similar dimension with " pendulum ", i.e. rope
Repeating motion.The free end so " put ", which increases local weight, can enhance the movement energy of pendulum;
Rope:(rope upper end [start] is, fixed) ∧ (rope lower end [end], be on-fixed) ∧ (rope lower ends
[end], adds, weight) ∧ (weight is independent heavy wisp) → (rope becomes, swing rod)
(weight is the wisp that independent heavy easy Bundles is tied up)
=(weight is object) ∧ (weight is, independent) ∧ (weight is, heavy) ∧
(weight be, what easy Bundles was tied up) ∧ (weight is, small)
C, chair, Gestalt structural dimension, which includes it, independence (thus being independent weight), it has bearing function,
Load in other words, and Te Dian are tied up, but scale is larger with easy Bundles;
Chair:(chair is object) ∧ (chair is, independent) ∧ (chair is that proportion is small) ∧
(chair be, what non-easy Bundles was tied up) ∧ (chair is, big)
D, pliers, Gestalt structural dimension include small scale, easily contact (i.e. easy Bundles is tied up), independence (individual object
Body), unit ratio it is great, etc.;
Pliers:(pliers is object) ∧ (pliers is, independent) ∧ (pliers is, heavy) ∧
(pliers be, what easy Bundles was tied up) ∧ (pliers is, small)
E, the quantity between the quantitative relation of the fixed position of two ropes as well as the Gestalt structural dimension of " pendulum "
Relationship --- the quantitative relation of i.e. two free ends.When people is as contact object, the spatial position quantitative relation between three is special
It is not the quantitative relation when the arm of people is as union body.When people pulls the free end of a rope and links another rope,
Space length cannot still meet Interface condition (quantity), it is therefore desirable to make another rope be in swing state and it is opposite when
Can just it link (San Dian spatial positions quantitatively meet Interface condition at this time), but another rope free end be not if increased office
Portion's weight, the then amplitude peak swung cannot reach requirement, therefore need to increase local weight (object).
Rope is coupled:(rope 1 is coupled, rope 2) ← (1 lower end of rope is coupled, 2 lower end of rope)
(1 lower end of rope and 2 lower end of rope, do, in opposite directions but (1 lower end of rope and 2 lower end of rope, have, level interval) ∧
On to movement) → (1 lower end of rope and 2 lower end of rope, do not have, level interval) → (rope 1 is coupled, rope 2)
2) it sets each factor correlation to compare, compares optimal correlation.As pliers than chair is more suitable for and restricts
Son is combined into " pendulum ".This is related to the restriction for the Gestalt structural dimension and its amount " put " and defining for optimization.
As the pendulum of " pendulum ", defined according to the Gestalt structural dimension of " pendulum " and its restriction of amount with what is optimized, respectively
It calculates and compares pliers and " applicable " magnitude in " pendulum " of chair.It obtains
The value of the value > chairs of pliers
3) link process, refer here to chronological order with limiting.As first by pliers and another rope free end connection group
It synthesizes " pendulum ", then people swings it, then descendant holds previous rope, and the other hand catches the lower end of " pendulum ", most
Afterwards, pliers is removed to link up Two root ropes.
(1 lower end of rope and 2 lower end of rope, do, mutually upwards to movement) ← | (pliers, connection, 1 lower end of rope) ∧ (people,
Make ... to swing, rope 1) or ((people makes, rope 1) ∧ (rope swings, gets up)) | ∧ | (people on the other hand, catches, rope 2)
∧ (people's the other hand, catches, the pliers of 1 lower end of rope) | → (people removes, pliers) ∧ (people, connection, 1 lower end of rope and rope 2
Lower end)
(pliers, connection, 1 lower end of rope) → (rope 1, becomes, pendulum)
Extensive pattern:
Actor x:| (x, can be with, activity) → (x, can be with mobile) ∧ (x, can to grab) | ∧ (x is, independent) ∧
(x is object) ∧ ...
The object y of the big softness of slenderness ratio:(y one end [start] is, fixed) ∧ (y one end [end], be on-fixed) ∧
(the y other ends [end], add, weight) ∧ (weight is independent heavy wisp) → (y becomes, swing rod)
Change the intermediary z of original state:(z is object) ∧ (z is, independent) ∧ (z is that proportion is small) ∧
(z be, what non-easy Bundles was tied up) (z is scale of construction t) to ∧
Y is coupled:(y1 is coupled, y2) ← (y1 movable ends are coupled, y2 movable ends)
But (y1 movable ends with y2 movable ends, have, level interval) ∧ (y1 movable ends and y2 movable ends, do, mutually upwards to
It is mobile) → (y1 movable ends and y2 movable ends, do not have, level interval) → (y1 is coupled, y2)
Calculating ratio pair:As the pendulum of " pendulum ", according to the Gestalt structural dimension of " pendulum " and its restriction of amount and optimization
It defines, calculate respectively and compares pliers and " applicable " magnitude in " pendulum " of chair.Obtain the value of z and the person of being maximized zj.
Behavior (connection):(y1 movable ends and y2 movable ends, do, mutually upwards to movement) ← | (z, connection, y1 movable ends) ∧
(people makes ... to swing, y1) or ((people makes, y1) ∧ (y1 swings, gets up)) | ∧ | (people on the other hand, catches, y2) (people is another by ∧
On the other hand, catch, the z of y1 movable ends) | → (people removes, z) ∧ (people, connection, y1 movable ends and y2 movable ends)
It can also be more abstractively extensive.
With IF THEN production system regular expression:
IF actors, THEN x
IF x, THEN (can be with activity)
IF can be with THEN activities
IF (can with, activity), THEN (can be with mobile) ∧ (can to grab) ∧ (...)
The ∧ objects of IF x, THEN independences
The object y of the big softness of slenderness ratio:(y one end [start] is, fixed) ∧ (y one end [end], be on-fixed) ∧
(the y other ends [end], add, weight) ∧ (weight is independent heavy wisp) → (y becomes, swing rod)
The object of the big softness of IF slenderness ratios, THEN y
IF y, THEN (y one end [start] is, fixed) ∧ (y one end [end], be on-fixed) ∧ (y other ends
[end], adds, weight) ∧ (weight is independent heavy wisp) → (y becomes, swing rod)
IF y, THEN (y one end [start] is, fixed)
IF y one end [start], THEN is fixed
IF y, THEN (y one end [end], be on-fixed)
IF y one end [end], THEN on-fixeds
IF y one end [end], THEN (the y other ends [end], add, weight)
IF y one end [end], THEN (add, weight)
IF adds, THEN weight
IF weight, the heavy wisp of THEN independences
IF (the y other ends [end], add, weight), THEN (y becomes, swing rod)
IF y, THEN (become, swing rod)
IF becomes, THEN swing rods
Y=(becoming, swing rod)/* equivalence relations */
=become/* equivalence relations */
=swing rod/* equivalence relations */
……
Directly natural language is compiled like predicate calculus form Yu Ju Zhuan Change for production system rule set with PROLOG program
Journey
If we want using a building blocks on another building blocks (above) concept.It is drilled with predicate
It calculates, we can recursively define above with two following formula with item o n: It is defined with these,
Following PROLOG program can be used to prove that for B when on C, A is on C when A is on B:
1):-Above(A,C)
2)On(A,B):-
3)On(B,C):-
4)Above(x,y):-On(x,y)
5)Above(x,y):-On(x,z),Above(z,y)
First possible target sums up generates fresh target (with rule 4):-On(A,C).Target fails, therefore we return
It traces back and sums up (with rule 5) one fresh target of generation to initial target examination is next:- On (A, z), Above (z, C).Sum up this
First word in target generates fresh target (with the fact 2):-Above(B,C).This target sums up with rule 4 to be generated:On
(B, C), it and the fact 3 sum up generation null object, and process terminates.Target:The generation of-Above (B, C) is considered to same
The recursive call of one program.
It is represented with natural language like predicate calculus form, quantifier can be saved, and directly turned by most simple thoughtcast collection
Become predicate form and weave into Prolog programs automatically:
A is on B, and for B when on C, A is on C:
1):- Above (A, C)=(A, Above, C)
2) On (A, B)=(A, On, B):-
3) On (B, C)=(B, On, C):-
4) Above (x, y)=(x, Above, y):- On (x, y)=(x, On, y)
5) Above (x, y)=(x, Above, y):- On (x, z), Above (z, y)=(x, 25On, y), (z, Above, y)
【Programming Skills】
Following instruction is compiled into program:
If GDP increases, then buys equity share.Otherwise, equity share is sold.
1. natural language is turned Change like predicate calculus form, the key clause collection merging of the most simple thoughtcast of acquisition turns Change and is
Production system rule set;
If GDP increases, then buys equity share.Otherwise, equity share is sold
=| IF (GDP increases) THEN (..., buy, equity share) | ∧ else (..., it sells, commonly
Stock)
IF (GDP, increase) THEN buys/and * rule meanings are:IF (GDP increases)
It is true, it is true that THEN, which is bought,.*/
IF (GDP, increase) THEN equity shares/* rule meanings are:(GDP increases IF
It is long) it is true, it is true that THEN, which (buys) equity share,.*/
IF (GDP increases) THEN (GDP increases)
IF (GDP increases) THEN GDPs
IF (GDP increases) THEN increases
IF GDPs THEN growths/* rule meanings are:IF GDPs are true, and THEN growths are true.*/
IF buys THEN equity shares/* rule meanings:It is true that IF, which is bought, and it is true that THEN, which (buys) equity share,.*/
IFelse THEN (..., sell, equity share)
IF sells THEN equity shares
If=if or while, then=then or do, otherwise=else
2. it is reached with pseudo table
For those skilled in the art, technical solution that can be as described above and design are made other each
The corresponding change of kind and deformation, and all these changes and deformation should all belong to the protection model of the claims in the present invention
Within enclosing.
Claims (10)
1. a kind of natural language production system of intelligent body, which is characterized in that including:
Input unit, for obtaining nature language statement;
Cutting unit, for natural language sentence to be obtained most simple thoughtcast clause like predicate calculus form by natural sentence
Collection;
Converting unit represents conversion for most simple thoughtcast clause set to be carried out production system;
Unit, for transformed sentence to be carried out creative thinking study;
Output unit, for intelligent body to be controlled to make corresponding actions;Or generated statement output is as conclusion;Or as learning outcome
It is stored.
2. the natural language production system of intelligent body according to claim 1, which is characterized in that the cutting unit leads to
It crosses sentence cutting and is converted to and with different levels gathered by three most simple thoughtcast clauses formed.
3. the natural language production system of intelligent body according to claim 1, which is characterized in that in the cutting unit
It establishes that sentence is bunched and sentence is bunched outer knowledge item, using clause as pointer, indefinite concept is made in sentence outer knowledge item of bunching
Supplement retrieval.
4. the natural language production system of intelligent body according to claim 1, which is characterized in that the converting unit profit
Most simple thoughtcast clause set is carried out production system with the equivalence relation in natural language sentence to represent to convert, including:It is right
The production system of knowledge and relationship is represented, is represented with the production system of reasoning, to learning the production with planning expressing
The new sentence of system representation, natural language and the production system of phrase generation represent.
5. the natural language production system of intelligent body according to claim 1, which is characterized in that the output unit root
Self-programing is carried out according to transformed sentence, for intelligent body itself to be controlled to act.
6. a kind of natural language production method of intelligent body, which is characterized in that include the following steps:
S1, nature language statement is obtained;
S2, natural language sentence is obtained into most simple thoughtcast clause set by natural sentence like predicate calculus form;
S3, most simple thoughtcast clause set is carried out to production system expression conversion;
S4, transformed sentence is subjected to creative thinking study;
S5, control intelligent body make corresponding actions;Or generated statement output is as conclusion;Or it is stored as learning outcome.
7. the natural language production method of intelligent body according to claim 6, which is characterized in that the step S2 passes through
Sentence cutting is converted to different levels by the three most simple thoughtcast clauses formed set.
8. the natural language production method of intelligent body according to claim 6, which is characterized in that established in the step S2
Outer knowledge item that sentence is bunched and sentence is bunched using clause as pointer, supplements indefinite concept in sentence outer knowledge item of bunching
Retrieval.
9. the natural language production method of intelligent body according to claim 6, which is characterized in that the step S3 is utilized certainly
Most simple thoughtcast clause set is carried out production system and represents conversion by the equivalence relation in right language statement, including:To knowledge
It represents with the production system of relationship, represented with the production system of reasoning, to learning the production system with planning expressing
It represents, the production system that the new sentence of natural language and phrase generate represents.
10. the natural language production method of intelligent body according to claim 6, which is characterized in that the step S5 roots
Self-programing is carried out according to transformed sentence, for intelligent body itself to be controlled to act.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810073361.7A CN108255814A (en) | 2018-01-25 | 2018-01-25 | The natural language production system and method for a kind of intelligent body |
PCT/CN2018/118049 WO2019144699A1 (en) | 2018-01-25 | 2018-11-29 | Natural language production system and method for intelligent agent |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810073361.7A CN108255814A (en) | 2018-01-25 | 2018-01-25 | The natural language production system and method for a kind of intelligent body |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108255814A true CN108255814A (en) | 2018-07-06 |
Family
ID=62742605
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810073361.7A Pending CN108255814A (en) | 2018-01-25 | 2018-01-25 | The natural language production system and method for a kind of intelligent body |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN108255814A (en) |
WO (1) | WO2019144699A1 (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109241531A (en) * | 2018-08-30 | 2019-01-18 | 王立山 | The learning method and system of natural language mind over machine |
CN109800344A (en) * | 2019-01-28 | 2019-05-24 | 王立山 | A kind of automatic programming method and its system of natural language mind over machine |
WO2019144699A1 (en) * | 2018-01-25 | 2019-08-01 | 王立山 | Natural language production system and method for intelligent agent |
CN117149163A (en) * | 2019-01-10 | 2023-12-01 | 布莱恩认知私人有限公司 | Natural solution language |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103853824A (en) * | 2014-03-03 | 2014-06-11 | 沈之锐 | In-text advertisement releasing method and system based on deep semantic mining |
CN106055537A (en) * | 2016-05-23 | 2016-10-26 | 王立山 | Natural language machine recognition method and system |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102622342B (en) * | 2011-01-28 | 2018-09-28 | 上海肇通信息技术有限公司 | Intermediate family of languages system, intermediate language engine, intermediate language translation system and correlation method |
CN108255814A (en) * | 2018-01-25 | 2018-07-06 | 王立山 | The natural language production system and method for a kind of intelligent body |
-
2018
- 2018-01-25 CN CN201810073361.7A patent/CN108255814A/en active Pending
- 2018-11-29 WO PCT/CN2018/118049 patent/WO2019144699A1/en active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103853824A (en) * | 2014-03-03 | 2014-06-11 | 沈之锐 | In-text advertisement releasing method and system based on deep semantic mining |
CN106055537A (en) * | 2016-05-23 | 2016-10-26 | 王立山 | Natural language machine recognition method and system |
Non-Patent Citations (1)
Title |
---|
陈少强 等: "一种产生式系统的扩展方法", 《江西师范大学续保(自然科学版)》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019144699A1 (en) * | 2018-01-25 | 2019-08-01 | 王立山 | Natural language production system and method for intelligent agent |
CN109241531A (en) * | 2018-08-30 | 2019-01-18 | 王立山 | The learning method and system of natural language mind over machine |
CN117149163A (en) * | 2019-01-10 | 2023-12-01 | 布莱恩认知私人有限公司 | Natural solution language |
CN109800344A (en) * | 2019-01-28 | 2019-05-24 | 王立山 | A kind of automatic programming method and its system of natural language mind over machine |
Also Published As
Publication number | Publication date |
---|---|
WO2019144699A1 (en) | 2019-08-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Levy | In the plex: How Google thinks, works, and shapes our lives | |
CN106055537B (en) | A kind of natural language machine identification method and system | |
CN108255814A (en) | The natural language production system and method for a kind of intelligent body | |
Visser | Sahu-Indonesian-English Dictionary | |
Rangan et al. | Environmental history of botanical exchanges in the Indian Ocean world | |
Borowy et al. | History of the Future of Economic Growth | |
Williams et al. | Architecture and Mathematics from Antiquity to the Future | |
CN107436942A (en) | Word embedding grammar, system, terminal device and storage medium based on social media | |
Kong et al. | Utilizing a crowdsourced phrasal lexicon to identify cultural ecosystem services in El Cajas National Park, Ecuador | |
Kodalak | Spinoza and Architecture | |
Grosholz | Great circles: the transits of mathematics and poetry | |
Zhang et al. | Learning to transform service instructions into actions with reinforcement learning and knowledge base | |
Nygren et al. | The status quo of digital humanities in Sweden: past, present and future of digital history | |
CN104951096A (en) | Chinese character coding input method for coordinate shape codes of eight categories of stroke shapes | |
Fawcett | The cultural classification of ‘things’ | |
Ramamurthy et al. | NLPGym--A toolkit for evaluating RL agents on Natural Language Processing Tasks | |
Han et al. | LLMs as Visual Explainers: Advancing Image Classification with Evolving Visual Descriptions | |
Paulik | Computer simulation models for fisheries research, management, and teaching | |
CN112463988A (en) | Method for extracting Chinese classical garden information | |
da Silva Sinha | Linguistic and cultural conceptualisations of Time in Huni Kuĩ, Awetý and Kamaiurá indigenous communities of Brazil | |
Hulsewâe | Thought and Law in Qin and Han China: Studies Dedicated to Anthony Hulsewâe on the Occasion of His Eightieth Birthday | |
Hurford | Artificially growing a numeral system | |
Al-Azhari | Design by shape grammar precedents as an experiential learning module of residential architecture of Amman city | |
Tadmor et al. | Borrowability and the notion of basic vocabulary | |
Bhattacharya | Origin and Evolution of Theatre: A Biological Interpretation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180706 |