For people who want to constantly improve, learn something and constantly learn something new, we have specially made this category. It contains exclusively educational, useful content that you will definitely enjoy. A large number of videos, perhaps, can even compete with the education that we are given at school, college or university. The biggest advantage of instructional videos is that they try to provide the latest, most up-to-date information. The world around us in the era of technology is constantly changing, and printed educational publications simply do not have time to give out fresh information.
Among the videos you can also find educational videos for preschool children. There your child will be taught letters, numbers, counting, reading, etc. Agree, a very good alternative to cartoons. For primary school students, you can also find English language training, help in studying school subjects. For older students, training videos have been created that will help prepare for tests, exams, or simply deepen their knowledge in a particular subject. The acquired knowledge can qualitatively affect their mental potential, as well as please you with excellent grades.
For young people who are out of school, in university or not, there are plenty of entertaining educational videos available. They can help them deepen their knowledge of the profession for which they are studying. Or get a profession, such as a programmer, web designer, SEO optimizer, and so on. Universities do not teach such a profession yet, so you can become a specialist in this advanced and relevant area only by doing self-education, which is what we try to help by collecting the most useful videos.
For adults, this topic is also relevant, since it often happens that after working in the profession for years, it comes to the understanding that this is not yours and you want to learn something more suitable for yourself and at the same time profitable. Also among this category of people there are often videos on the type of self-improvement, saving time and money, optimizing their lives, in which they find ways to live much better and happier. Even for adults, the topic of creating and developing your own business is very well suited.
Also among the educational videos there are videos with a general focus, which are suitable for almost any age, in which you can learn about how life originated, what theories of evolution exist, facts from history, etc. They perfectly expand the horizons of a person, make him a much more erudite and pleasant intellectual interlocutor. It is really useful to watch such informative videos for everyone without exception, since knowledge is power. We wish you a pleasant and useful viewing!
In our time, it is simply necessary to be what is called "on the wave." This refers not only to news, but also to the development of one's own mind. If you want to develop, explore the world, be in demand in society and interesting, then this section is for you.
5.1. Differences between knowledge and data
A characteristic feature of intelligent systems is the availability of knowledge necessary to solve problems in a particular subject area. In this case, a natural question arises, what is knowledge and how does it differ from ordinary data processed by a computer.
Data is information of an actual nature that describes objects, processes and phenomena of the subject area, as well as their properties. In the processes of computer processing, data goes through the following stages of transformation:
The original form of data existence (results of observations and measurements, tables, directories, charts, graphs, etc.);
Presentation in special languages of the description of data intended for input and processing of initial data in a computer;
Databases on machine storage media.
Knowledge is a more complex category of information than data. Knowledge describes not only individual facts, but also the relationships between them, which is why knowledge is sometimes called structured data. Knowledge can be obtained through the processing of empirical data. They are the result of a person's mental activity aimed at generalizing his experience gained as a result of practical activities.
In order to endow IIS with knowledge, they must be presented in a certain form. There are two main ways of imparting knowledge to software systems. The first is to put the knowledge into a program written in a conventional programming language. Such a system will be a single program code in which knowledge is not placed in a separate category. Despite the fact that the main task will be solved, in this case it is difficult to assess the role of knowledge and understand how it is used in the process of solving problems. It is not easy to modify and maintain such programs, and the problem of replenishing knowledge can become insoluble.
The second method is based on the concept of databases and consists in placing knowledge into a separate category, i.e. knowledge is presented in a certain format and placed in the knowledge base. The knowledge base is easily updated and modified. It is an autonomous part of an intelligent system, although the inference mechanism implemented in the logical block, as well as the means of conducting a dialogue, impose certain restrictions on the structure of the knowledge base and operations with it. This method is adopted in modern IIS.
It should be noted that in order to put knowledge into a computer, it must be represented by certain data structures corresponding to the chosen development environment of an intelligent system. Consequently, in the development of IIS, knowledge is first accumulated and represented, and at this stage, human participation is mandatory, and then knowledge is represented by certain data structures that are convenient for storage and processing in a computer. Knowledge in IIS exists in the following forms:
Initial knowledge (rules derived from practical experience, mathematical and empirical relationships that reflect the mutual relationships between facts; patterns and trends that describe the change in facts over time; functions, diagrams, graphs, etc.);
Description of the initial knowledge by means of the chosen knowledge representation model (a set of logical formulas or production rules, a semantic network, frames, etc.);
Representation of knowledge by data structures that are intended for storage and processing in a computer;
Knowledge bases on machine storage media.
What is knowledge? Let's give some definitions.
From the explanatory dictionary of S. I. Ozhegov: 1) “Knowledge is the comprehension of reality by consciousness, science”; 2) "Knowledge is a set of information, knowledge in any area."
The definition of the term "knowledge" includes mostly philosophical elements. For example, knowledge is a practice-tested result of cognition of reality, its correct reflection in the human mind.
Knowledge is the result obtained by cognition of the surrounding world and its objects. In the simplest situations, knowledge is considered as a statement of facts and their description.
AI researchers are giving more specific definitions of knowledge.
"Knowledge is the laws of the subject area (principles, connections, laws) obtained as a result of practical activities and professional experience, allowing specialists to set and solve problems in this area" .
"Knowledge is well-structured data, or data about data, or metadata".
"Knowledge is formalized information that is referred to or used in the process of inference".
In the field of AI systems and knowledge engineering, the definition of knowledge is linked to inference: knowledge is information on the basis of which the inference process is implemented, i.e. Based on this information, various conclusions can be drawn from the data available in the system using inference. The inference mechanism allows you to link together separate fragments, and then draw a conclusion on this sequence of connected fragments.
Knowledge is formalized information that is referred to or used in the process of inference (Fig. 5.1.).
Rice. 5.1. The process of inference in IS
By knowledge we mean the totality of facts and rules. The concept of a rule representing a fragment of knowledge has the form:
If<условие>then<действие>.
This definition is a special case of the previous definition.
However, it is recognized that the distinctive qualitative features of knowledge are due to their great opportunities in the direction of structuring and interconnectedness of constituent units, their interpretability, the presence of metrics, functional integrity, and activity.
There are many classifications of knowledge. As a rule, with the help of classifications, knowledge of specific subject areas is systematized. At the abstract level of consideration, we can talk about the features by which knowledge is divided, and not about classifications. By its nature, knowledge can be divided into declarative and procedural.
Declarative knowledge is a description of facts and phenomena, fixes the presence or absence of such facts, and also includes descriptions of the main connections and patterns in which these facts and phenomena are included.
Procedural knowledge is a description of the actions that are possible when manipulating facts and phenomena to achieve the intended goals.
To describe knowledge at the abstract level, special languages have been developed - knowledge description languages. These languages are also divided into languages of procedural type and declarative type. All knowledge description languages focused on the use of traditional von Neumann architecture computers are procedural languages. The development of declarative languages suitable for knowledge representation is an actual problem of today.
According to the method of acquiring knowledge, it can be divided into facts and heuristics (rules that allow you to make a choice in the absence of exact theoretical justifications). The first category of knowledge usually points to well-known circumstances in the subject area. The second category of knowledge is based on the own experience of an expert working in a specific subject area, accumulated as a result of many years of practice.
According to the type of presentation, knowledge is divided into facts and rules. Facts are knowledge of the “A is A” type, such knowledge is typical for databases and network models. Rules, or products, are knowledge of the "IF A, THEN B" type.
In addition to facts and rules, there is also meta-knowledge - knowledge about knowledge. They are necessary for knowledge management and for efficient organization of inference procedures.
The form of knowledge representation has a significant impact on the characteristics of IMS. Knowledge bases are models of human knowledge. However, all the knowledge that a person attracts in the process of solving complex problems cannot be modeled. Therefore, in intelligent systems, it is required to clearly separate knowledge into those that are intended for processing by a computer, and knowledge used by a person. Obviously, in order to solve complex problems, the knowledge base must have a sufficiently large volume, and therefore problems of managing such a database inevitably arise. Therefore, when choosing a knowledge representation model, factors such as uniformity of representation and ease of understanding should be taken into account. The homogeneity of the presentation leads to a simplification of the knowledge management mechanism. Ease of understanding is important for users of intelligent systems and experts whose knowledge is embedded in IIS. If the form of knowledge representation is difficult to understand, then the processes of acquiring and interpreting knowledge become more complicated. It should be noted that it is quite difficult to simultaneously meet these requirements, especially in large systems, where structuring and modular representation of knowledge becomes inevitable.
Solving the problems of knowledge engineering puts forward the problem of converting information received from experts in the form of facts and rules for their use into a form that can be effectively implemented in the machine processing of this information. For this purpose, various models of knowledge representation have been created and used in existing systems.
The classical models of knowledge representation include logical, production, frame and semantic network models.
Each model has its own knowledge representation language. However, in practice, it is rarely possible to get by with the framework of one model when developing IIS, except for the simplest cases, so the representation of knowledge turns out to be difficult. In addition to the combined representation using various models, special tools are usually used to reflect the features of specific knowledge about the subject area, as well as various ways to eliminate and take into account the fuzziness and incompleteness of knowledge.
Data and knowledge. Basic definitions.
The information that computers deal with is divided into procedural and declarative. Procedural information is embodied in programs that are executed in the process of solving problems, declarative information - in the data with which these programs work.
The standard form of information representation in a computer is a machine word, consisting of a number of binary digits, or bits, defined for a given type of computer. A machine word for representing data and a machine word for representing instructions that make up a program may have the same or different number of bits. The same number of digits in machine words for commands and data allows them to be considered in the computer as the same information units and to perform operations on commands as on data. The contents of the memory form the information base. The machine word is the main characteristic of the infobase, because its length is such that each machine word is stored in one standard memory cell, provided with an individual name - the address of the cell. By this name, information units are retrieved from the computer memory and written to it. High-level programming languages use abstract data types whose structure is specified by the programmer.
The emergence of databases (DB) marked another step towards the organization of work with declarative information. Databases can simultaneously store large amounts of information, and special tools that form a database management system (DBMS) allow you to effectively manipulate data, if necessary, extract them from the database and write them in the right order to the database.
With the development of IP research, the concept of knowledge has emerged, which combines many features of procedural and declarative information. In a computer, knowledge, just like data, is displayed in symbolic form - in the form of formulas, text, files, information arrays, etc. Therefore, we can say that knowledge is data organized in a special way. In AI systems, knowledge is the main object of formation, processing and research. The knowledge base, along with the database, is a necessary component of the AI software package. Machines that implement AI algorithms are called knowledge-based machines, and a subsection of AI theory related to the construction of expert systems is called knowledge engineering.
Differences between data and knowledge:
1. internal interpretability of knowledge (for example: data - 243849..., knowledge - natural language sentences).
2. knowledge activity. If there is knowledge, then the emergence of new knowledge can lead to a change in old knowledge and the emergence of new ones.
3. connectedness of knowledge. Knowledge is not interesting in itself, it is interesting in the aggregate (knowledge system).
4. Knowledge is dynamic, while data is usually static.
Intensional knowledge is defined through the concept of a higher level with indication of specific properties. Extensional knowledge is defined in terms of lower level concepts, usually by simply enumerating them. As a rule, extensions are stored in databases, and intensions in knowledge bases. Knowledge according to the method of presentation is divided into declarative (information is described) and procedural (recorded in the algorithm). The main direction of movement in the field of knowledge representation is the greater use of declarative knowledge.
Classifications of knowledge and their models
There are many ways to classify knowledge. Let us dwell on the classification according to the carrier of knowledge. Knowledge is divided into:
1. Formalized
reference guides,
encyclopedias,
Knowledge of corporate information systems
2. Personal
Craft related skills
sports skills,
ways of thinking, analyzing,
ways of doing work
Formalized knowledge is usually already placed on tangible media - books, brochures, Internet / Intranet sites, data files, CIS (ERP). These ways of organizing knowledge are very good and time-tested. We are unlikely to be able to significantly improve them to reflect on the performance or other economic indicators of your organization.
Personal knowledge, on the other hand, is usually held only in the minds of its bearers. In order to make them the property of the organization, it is necessary that knowledge be actively transferred between employees. For this, mentoring and internal corporate training systems have long existed.
A lot of personal knowledge can be formalized. This concerns, first of all, the methods and ways of performing work that are accepted and optimal in your organization. Depending on the stage of development of the organization, work methods evolve from creative, created in the workplace through trial and error to best practices in the industry, enshrined in business process documentation, ERP system and organization policy.
Frame definitions. The frame as a list of properties and as a network. Hierarchy and property inheritance
A frame is a certain structure of knowledge representation, which, when filled with appropriate values, turns into a description of a specific factor, event or situation. A frame is the minimum possible description of the essence of any phenomenon, event, situation, process or object. Minimality means that with further simplification of the description, its completeness is lost, it ceases to determine the unit of knowledge for which it is intended. The frame has a certain structure, consisting of many elements - slots. Each slot, in turn, is represented by a specific data structure, procedure, or may be associated with another frame. The frame structure can be represented as follows:
FRAME NAME: (1st slot name: 1st slot value), (2nd slot name: 2nd slot value), ... (Nth slot name: Nth slot value).
We present the same record in the form of a table, supplemented by two columns.
The value of a slot can be the name of another frame; in this way they form networks of frames consisting of selected vertices and links. The top level of the frame represents the corresponding concepts, and the subsequent levels are terminal slots that contain specific values. The hierarchy of objects is implemented through the property research apparatus, when classes of objects of a certain level inherit the structure of frame classes of a higher level. If the object, cat. is described by a certain group of frames is in a conceptual relationship with the upper and lower levels of frames, then resp. to him, frames are constructed taking into account hierarchical relationships and, at the same time, inheritance of properties impl. through slots or frames with the same name.
Data and knowledge
Information
Data
procedural declarative
Subject area
Knowledge
inference
facts Heuristics
withdrawal mechanism, inference or output machine.
interface
Knowledge base,
withdrawal mechanism,
User interface.
The concept of a formal system
The basis of logical models is the concept of a formal system defined by a quadruple M = (T, P, A, F).
Lots of T there are many basic elements of different nature, for example, words from some limited vocabulary. It is assumed that there is a procedure P( T) checking whether an arbitrary element belongs to a set T.
Lots of P there are many syntax rules. With their help from the elements T form syntactically correct expressions, for example, syntactically correct expressions are built from the words of a limited dictionary. There must be a procedure P( R) to determine whether
some expression is syntactically correct.
in multitude R a subset is allocated BUT a priori true expressions (axioms). There must be a procedure P( BUT) checking whether any syntactically correct expression belongs to the set BUT.
Lots of F there are many semantic inference rules. Applying them to elements BUT, you can get new syntactically correct expressions, to which you can again apply the rules from F. This is how it is formed set of inferred in this formal system of expressions. If there is a procedure P( F), which makes it possible to determine for any syntactically correct expression whether it is derivable, then the corresponding formal system is called solvable.
For the knowledge included in the knowledge base, we can assume that the set BUT form all information units entered into the knowledge base, and with the help of inference rules, new ones are derived from them derived knowledge. In other words, the formal system is a generator of new knowledge that forms a set withdrawn in this system knowledge.
This model underlies the construction of many deductive IIS. In such systems, the knowledge base is described in the form of sentences and axioms of the theory, and the inference mechanism implements the rules for constructing new sentences from the knowledge base. The input of the system is a description of the problem in the language of this theory in the form of a request (suggestion, theorem), which is not explicitly presented in the KB. The process of operation of the inference mechanism is called the proof of the query (theorem).
The use of logics of various types in the construction of syntactic and semantic rules generates logical models of various types.
propositional calculus
The propositional calculus studies sentences that can be either true or false. Not every sentence is a statement. For example, it is meaningless to talk about the truth of interrogative sentences. Sentences are not statements for which there is no consensus about whether these sentences are true or false. Apparently, not everyone will agree with the statement "mathematical logic is a fascinating subject."
The sentence "It was snowing" is also not a statement, since in order to judge its truth, additional information is needed about when and where it snowed.
Combining sentences using connectives like "And", "or",“if…then…”, you can form new sentences.
The propositional calculus uses five logical connectives: negation, conjunction, disjunction, implication, and equivalence.
conjunction (logical AND) is true only if both of its constituent statements are true.
Disjunction (logical OR) is false only if both of its components are false.
The implication (corresponds to the link " If...then...”), the first operand is called the premise, and the second operand is called the conclusion. An implication is false only if its premise is true and its conclusion is false.
Boolean operation equivalence corresponds to the link " then and only then". Its result is true if both statements are either both true or both false.
Boolean negation performed on a single statement. A proposition and its negation always have opposite truth values.
The symbols used to denote statements are called atoms.
Well-formed formulas in propositional logic are recursively defined as follows:
1) an atom is a formula;
2) if A And B are formulas, then the formulas are
and Ø A, A Ù B, A Ú B, A ® B, A « B.
Here the links are symbolized:
Ú - logical OR(disjunction);
u - boolean AND(conjunction);
® - boolean SHOULD(implication);
"- logical EQUIVALENTLY(equivalence);
Ø - logical negation.
interpretation formula is called assigning to each atom included in the formula a truth value ( true or False).
Formula consisting of n different atoms, has 2 n various interpretations.
A formula that is true under all interpretations is called generally valid(for example, A Ú Ø A).
A formula that is false under all interpretations is called controversial(for example, A ÙØ A).
A formula for which there is at least one interpretation under which it is true is called doable.
equivalent are called formulas whose truth values are the same under all interpretations. Formulas can be converted from one form to another using equivalent substitutions.
For transformations of propositional calculus formulas, the following equivalences are used:
1) A Ú Ø A = true(true);
A Ù Ø A = false(False);
2) the rule of double negation
Ø (Ø A) = A;
3) A ® B = Ø A Ú B;
4) A « B = (A ® B) Ù ( B ® A);
5) laws of commutativity
A Ú B = B Ú A, A Ù B = B Ù A;
6) laws of associativity
(A Ú B) Ú C =A Ú ( B Ú C), (A Ù B) Ù C = A Ù ( B Ù C);
7) distribution laws
A Ú ( B Ù C) = (A Ú B) Ù ( A Ú C), A Ù ( B Ú C) = (A Ù B) Ú ( A Ù C);
8) de Morgan's laws
Ø( A Ú B) = Ø A Ù Ø B, Ø( A Ù B) = Ø A Ú Ø B;
9) A ® B = Ø B ® Ø A.
Predicate calculus
The apparatus of propositional calculus in many cases does not allow a satisfactory description of the subject area. A significant part of subject areas can be described by means of first-order predicate calculus. To do this, consider:
a) constants denoting an individual object or concept;
b) variables that at different times can denote different objects;
c) terms, the simplest of which are constants and variables, and in a more general case, represented by expressions of the type , where is a functional symbol, and are terms;
d) predicates used to represent relationships between objects in some subject area;
e) quantifiers - a means of setting the quantitative characteristics of the subject area.
Predicate is a logical function that accepts only truth values true" or " False».
A predicate consists of a predicate symbol and the corresponding ordered set of terms that are its arguments. predicate symbol P used to name relationships between objects. If he has n arguments, it is called n-local predicate symbol.
The record, which is the simplest (atomic) formula, means that the statement is true: objects are related by relation P.
Using the same logical connectives as in propositional calculus ( AND, OR, NOT, SHOULD, EQUIVALENTLY), you can build more complex formulas.
To determine the scope of variables in formulas, quantifiers (universality) and (existence) are used. Quantifiers allow us to construct statements about a set of objects and formulate statements that are true for this set.
Predicate calculus formulas (PPF - well-formed formulas) are defined recursively as follows:
1. an atom is a formula;
2. if A And B are formulas, then the formulas are and
Ø A, A Ù B, A Ú B, A ® B, A « B;
3. if - is a formula, then the formulas are and and .
The interpretation of formulas in the predicate calculus is the assignment of ranges of values to all constants, function and predicate symbols. Formula interpreted on the area D, takes the values true or False according to the following rules:
a) if the values of the formulas are given A And B, then the truth values of the formulas Ø A, A Ù B, A Ú B, A ® B, A « B are obtained from truth tables valid for propositional calculus;
b) the formula gets a value true if for each of D has the meaning true, otherwise its value is False.
c) the formula gets a value true, if for at least one of D has the meaning true, otherwise its value is false.
Formula A eat logical consequence formulas if and only if for any interpretation in which the formula 
true formula A is also true.
In addition to the formulas for equivalent transformations given for the propositional calculus, the following are true in the predicate calculus:
Ø($ ) = () (Ø );
Ø() = () (Ø ).
Frame types
For educational purposes There are two types of frames: a prototype frame and an example frame. Frame - prototype reflects knowledge about abstract stereotypical concepts, which are classes of some specific objects. Prototype frames reflect intensional knowledge, i.e. generalized knowledge about the patterns inherent in the considered class of objects. Frames - Examples reflect knowledge about specific facts of the subject area, or the so-called extensional knowledge. The transition from the frame-prototype to the frame-instance is performed during the procedure of the value of the frame-prototype in the process of the inference mechanism.
As an example, consider a simplified frame diagram - a prototype of the DATE concept:
<ДАТА> (<МЕСЯЦ><имя>)(<ДЕНЬ><целые числа {1,2,…, 31}>)
(<ГОД><функция>)(<ДЕНЬ НЕДЕЛИ><перечень {ПНД,ВТР,…,ВСК}>
<функция>)
The name of the frame-prototype is DATE. In the MONTH slot, NAME is written in place of the value, i.e. the slot value can be any literal expression. The value of the DAY slot is integers, and a list of them is given in the slot. The YEAR slot contains a function that can implement the following actions. If the year is specified in the input sentence, then it is entered in the value field of the slot in the frame - example; if no year is specified, the missing value is filled in with the current year. These kinds of functions are called default functions.
The WEEK DAY slot also defines a function that, when processing an input message, will be called automatically to check for
inconsistency in the value of the day of the week specified by the user, or computed
this value if the user has not specified it.
Specific Frame - An example of a DATE frame might look like this:
<ISA DATE>(<МЕСЯЦ><ИЮНЬ>)(<ДЕНЬ><5>)
Label ISA means that this frame is an example frame. Only 2 slots are filled here. The values of the rest can be calculated using the appropriate procedures.
The procedures included in the slot are divided into two types: procedures - demons and
procedures are servants.
Procedures are demons are activated automatically every time the data enters the corresponding frame - example, or is removed from it. Thus, the procedure built into the WEEKDAY slot in the above example is a representative of the daemon procedure. With the help of procedures of this type, all routine operations related to maintaining databases and knowledge are performed.
Procedures are servants activated only upon request. An example of such a procedure is the function built into the YEAR slot in the DATE prototype frame, which is called only if the user has not specified a year.
Rice. 4.6 Frame network
on the frame Child. Inheritance of the slot "loves" from the frame Child.
Question 2. What is the age of the students?
Answer: 6-17 - the value of the "age" slot is taken from the Student frame. The value from the Child frame is not taken, because the meaning is explicitly stated in the "student" frame itself about which the question is being asked.
By function There are the following types of frames:
Frames are objects (example above);
Frames - operations (for example, the frame "process of synthesis of corrective devices", slots: model, calculation algorithm, parameters, etc.);
Frames - situations (for example, the frame "Emergency operation of the analog sensor", slots: voltage, current, etc.);
Frames - scenarios (for example, the frame "Extinguishing a fire", slots: a place of fire, extinguishing agents, etc.).
The frame model of knowledge representation is used in languages FRL(Frame Representation Language) ,KRL(Knowledge Representation Language) and etc.
Features of inference
In frame languages, the main operation is pattern search. A sample is a frame in which not all structural units are filled, but only those by which the necessary frames will be found among the frames stored in the system memory. The pattern might, for example, contain the name of a frame, as well as the name of some slot in the frame, followed by the value of the slot. Such a pattern checks for the presence in system memory of a frame with the given name and the given value of the slot specified in the pattern. The pattern may contain the name of some slot and its value. Then the pattern search procedure must ensure that all frames containing a slot with the same name and slot value as the pattern have been selected. Finally, some boolean function can be given on the frame name, some slot names, and some slot values. Thus, inference in a network of frames is based on a matching operation.
Other procedures typical for frame languages are procedures for filling slots with data, as well as procedures for introducing new prototype frames (i.e., new knowledge) into the system and introducing new relationships between them.
Consider a description fragment from the "world of blocks" (Fig. 4.7) in the form of frames in the FRL language.
Rice. 4.7 "World of Blocks"
(frame (name (cube)) (length (NULL)) (width (IF-DEFAULT (use length))) (height (IF-DEFAULT (use length)))) (f frame (name (B 1)) (AKO (cube)) (color (red)) (length(80))) (f frame (name (B 2)) (AKO (cube)) (color (green)) (length (65))))
Slot AKO indicates that the objects B 1 and B 2 are a subtype of the object cube and inherit its properties, namely, length = width = height. Procedure - daemon IF-DEFAULT fills in the default slot values.
Let's say the robot is given the order "Take the yellow object that supports the pyramid." In the knowledge representation language, the question is written as follows:
(object X (color (yellow)) (hold Y (type (pyramid))))
The pattern matching program finds the description of objects in the knowledge base:
(frame (name (B 3)) (type (block)) (color (yellow)) (size (20 20 20)) (coordinate (20 50 0)) (hold (P 2)))
(frame (name(P 2)) (type (pyramid)) ...)
Response received: X = B 3, Y = P 2, and the robot is given a command take(object B 3).
The advantages of frames as a knowledge representation model are the ability to structure the knowledge base due to the properties of hierarchy and inheritance. The disadvantage is the complexity of the organization of logical inference.
Lecture. Basics of building a production system
Applying metarules
Sometimes, in order to decide which rule to activate, it is desirable to use specific knowledge rather than follow a general conflict resolution strategy. To this end, some rule interpreters include tools that allow the programmer to formulate and introduce meta-rules into the program. Meta-rules define the rules by which the selection from the list of applications of those rules that should be considered in the first place or, moreover, be mandatory is performed.
Meta-rules allow you to significantly narrow the range of candidate rules based on some criterion or change the order of rule priorities. Meta-rules often use knowledge from a particular subject area. An example is the following meta-rule related to the system
topic of medical diagnostics MYCIN.
METARULE 001
IF (1) the infection belongs to the class pelvic abscess, And
(2) there are rules whose premise mentions
enterobacteria, And
(3) there are rules whose premise mentions
gram positive stain
TO with confidence 0.4 priority should be given to the first of the listed rules.
Lecture. Basic concepts in the field of artificial intelligence
The field of science, called "artificial intelligence", aims to identify the fundamental mechanisms underlying human activity in order to apply them in solving specific scientific and technical problems. "Reasonable" systems are created to work in environments where the presence of a person is impossible or life-threatening. These devices will have to operate in a wide variety of possible situations. It is impossible to describe these situations in advance with the degree of detail and unambiguity that would allow hard-coded behavior algorithms to be incorporated into the system being created. Therefore, systems armed with artificial intelligence should have adaptation mechanisms that would allow them to build programs of expedient activities to solve the tasks assigned to them based on the specific situation that is currently developing in their environment.
Such a formulation of the problem puts forward special tasks for researchers that have not previously arisen in the design of technical systems. These tasks include: description of a rich external environment and its reflection within the system (this task is called the task of knowledge representation); management of the knowledge bank, its replenishment, detection of contradictions and lack of knowledge; perception of the external environment with the help of various kinds of receptors (visual, tactile, auditory, etc.); understanding of natural language, which serves as a universal means of communication for a person; the perception of printed text and oral speech and the transformation of information contained in messages into a form of knowledge representation; activity planning is a task, the solution of which will allow the system to form plans for achieving the goal with the help of the means at its disposal; adaptation and learning based on experience.
This is the field of activity of specialists in the field of artificial intelligence systems. It lies at the intersection of a wide variety of disciplines: programming and psychology, technology and linguistics, mathematics and physiology.
So, the theory of artificial intelligence is the science of knowledge, how to extract it, represent it in artificial systems, process it inside the system and use it to solve practical problems. In other words, systems studied within the framework of artificial intelligence and created in line with this science are systems whose work is based on knowledge that reflects the semantics and pragmatics of the external world in which intelligent systems operate.
Thus, the main problems of artificial intelligence are the representation and processing of knowledge. The solution of these problems consists both in the development of effective models of knowledge representation, methods for obtaining new knowledge, and in the creation of programs and devices that implement these models and methods.
Elements of artificial intelligence are widely used to create intelligent computer software, automated control systems (ACS), design automation systems (CAD), information retrieval systems (IPS), database management systems (DBMS), expert systems (ES), systems decision support (DSS), i.e. allow to increase the level of intelligence of the created information systems.
Achievements in the field of artificial intelligence are used in industry (discovery and development of deposits, astronautics, automotive, chemistry, etc.), in the economy (finance, insurance, etc.), in the non-industrial sector (transport, medicine, communications etc.), in agriculture.
Artificial intelligence tools make it possible to develop models and programs for solving problems for which direct and reliable solution methods are not known. Such tasks belong to the field of human creative activity. Artificial intelligence specialists pose such scientific problems as proving mathematical theorems, diagnosing diseases or malfunctions in equipment, financial analysis of business entities, synthesis of programs based on specifications, understanding text in natural language, image analysis and identification of its content, robot control, etc.
Data and knowledge
Let us give definitions of the main concepts of the studied discipline and consider the differences between the concepts of "data" and "knowledge".
Information- a set of information perceived from the environment, issued to the environment or stored within the information system (IS).
Data- specific information presented in a formalized form about the objects of the subject area, their properties and relationships, reflecting events and situations in this area.
Data is presented in a form that allows automating their collection, storage and further processing. Data is a record of information in an appropriate form suitable for storing, transmitting, processing and obtaining new information.
The information that the computer deals with is divided into procedural and declarative.
procedural information is represented by programs that are executed in the process of solving problems, and declarative- the data that these programs process.
Any intellectual activity is based on knowledge of the subject area in which tasks are set and solved.
Subject area called a set of interrelated information necessary and sufficient to solve a certain set of problems. Knowledge about the subject area includes descriptions of objects, phenomena, facts, as well as relationships between them.
Knowledge- this is a generalized and formalized information about the properties and laws of the subject area, with the help of which the processes of solving problems, transforming data and knowledge themselves are implemented, and which is used in the process of inference.
inference is the generation of new statements (judgments) based on the initial facts, axioms and rules of inference.
Knowledge in terms of tasks to be solved in a certain subject area is divided into 2 large categories - facts and heuristics. Under facts usually understand well-known truths and circumstances in a given subject area. Heuristics are empirical algorithms based on informal considerations that limit the number of solutions and ensure the purposefulness of the behavior of the decision system, without guaranteeing, however, that the best solution is obtained. Such knowledge is based on the experience of a specialist (expert) in this subject area.
The concept of the procedure for obtaining solutions to problems (knowledge processing strategies) is associated with the knowledge of dough. In IIS, this procedure is called withdrawal mechanism, inference or output machine.
The knowledge with which the system works is stored in the knowledge base (KB).
To organize interaction with the IIS, it must have means of communication with the user, i.e. interface. The interface provides work with the KB and the output mechanism in a language of a sufficiently high level, close to the professional language of specialists in the subject area to which IIS belongs. In addition, the interface functions include support for the user's dialogue with the system, which allows the user to receive explanations of the system's actions, participate in the search for a solution to the problem, replenish and correct the knowledge base. Thus, the main parts of the IIS are:
Knowledge base,
withdrawal mechanism,
User interface.
Features of knowledge that distinguish them from data
Example. Let family ties act as the subject area. The objects of this subject area are such concepts as mother,
father, daughter, man, woman, etc.
Let the facts be known:
Victor is Tanya's father.
Vladimir is the father of Victor.
In Prolog, these facts are described as follows:
father (Victor, Tanya).
father (Vladimir, Victor).
Here "father" is a relation name or a predicate, and "victor", "tanya" and "vladimir" are constants.
Let be X, Y, Z– variables. Using variables X And Z, we can generally write the relation X is the father Z» in Prolog:
father ( X, Z).
Using the "father" predicate and variables X, Y, Z, we formulate a new relation "grandfather", namely:
If X is the father Z And
Z is the father Y
then X is a grandfather Y.
This form of writing the relationship "If ... Then" is called production rule, products or just rule.
In Prolog, the relation "grandfather" is written as follows:
grandfather ( X, Y): - father ( X, Z), father ( Z, Y).
The character ":-" is interpreted as "If".
On the example of the relation "grandfather", the general regularity of the definition of the concept of "grandfather" through the concept of "father" is formulated. The name "Vladimir", taken regardless of the relationship, does not indicate anything. Perhaps this is the name of a person or the name of a city. Numerical or other data is treated in the same way, for example, in a data file. The given, taken together with the relation, determines some meaning and thus constitutes knowledge.
Let us consider the features of knowledge, in which they differ from data.
1. Interpretability. The data stored in the computer memory can only be interpreted by the appropriate program. Data without a program does not carry any information, while knowledge has an interpretation, since it contains both data and their corresponding names, descriptions, relationships, i.e. along with data, information structures are presented that allow not only storing knowledge, but also using it.
Knowledge and information are important components of our life. These terms cannot be completely identified with each other. Consider what is meant by each of them and how knowledge differs from information.
Definition
Knowledge- systematized reliable ideas about objects and phenomena of reality. Knowledge is used by people to rationally organize their activities and solve emerging problems.
Information– information about concepts, facts, events, etc., in the transmission and acceptance of which people or special devices can participate. Animals communicate specific information to each other using signals. There is also genetic information passed from one organism to another.
Comparison
The fundamental factor that makes it possible to identify the difference between knowledge and information is that knowledge is acquired only through subjective comprehension. Information is independent and does not always reach the stage of awareness.
In the cognitive process, knowledge and information are at different levels. First, there is a perception of information transmitted by a certain source: a book, the Internet, a teacher ... After understanding, the information results in knowledge. The possessor of knowledge is able to play the role of a new source of information.
Thus, only information is transmitted and received, but knowledge cannot be transmitted. In order to become the owner of knowledge, it is necessary to perceive the necessary information and pass it through your own consciousness.
For example, a math teacher has knowledge in their subject area. Explaining to the class how to solve the problem, he does not directly transfer knowledge, but is a source of information. Students will be able to form knowledge only when they not only listen to the teacher, but also understand, realize what he is trying to convey to them.
Considering the difference between knowledge and information, it should be noted that there can be no excess of knowledge. After all, a person seeks to comprehend only what is really important and necessary for him. Information can come in abundance, people often feel a glut of it. Of the total amount of information, a small part is used to obtain knowledge.
It is knowledge that is the criterion of human education. After all, it is not enough just to get acquainted with the information - it is necessary to do a lot of mental work.
