ARTIFICIAL INTELLIGENCE Subject code:TCS-603

Submitted by:Chand a !he"ha #SIT$%

&NIT ': INTR()&CTI(N

A ti*icia+ inte++i,ence: Artificial Intelligence (AI) is the intelligence of machines and the branch of computer science which aims to create it.

AI research is highly technical and specialized, so much so that some critics decry the fragmentation of the field. Subfields of AI are organized around particular problems, the application of particular tools and around longstanding theoretical differences of opinion. The central problems of AI include such traits as reasoning, knowledge, planning, learning, communication, perception and the ability to mo e and manipulate ob!ects. "eneral intelligence is still a long#term goal of research. A ti*icia+#something that is man#made(not natural) Inte++i,ence#$atural intelligence %illitary intelligence Animal intelligence Artificial intelligence )e*inition:Intelligence may be defined as the characteristic that enable a being to generate thought based on either present en ironment or past e ent. Intelligence refers to the ability to store and sometimes later retrie e information relating to e&periences of a 'eing. Intelligence refers to the ability to recognise and recall past e ents when similar e ent take place with a iew to impro e performance ,such a recall being in an orderly and systematic manner. According to (ebsters $ew )ollegiate dictionary*Intelligence is the power of meeting any situationespecially and no el situation successfully by proper beha iour ad!ustments. $atural intelligence demands on retrie al of information either from e&ternal storage+memory or e&ternal en ironment caused by stimulus or a combination and path. Artificial tntelligence ,)omputer science test engineers with to design a thinking %achine to generate thought in a systematic manner. Some de*inition o* AI: -)AI is a branch of computer science which allows us to ac.uire represent and manipulate knowledge. /)AI is the study of mental features through the use of computational models. 0)AI is part of computer science concerned with designing intelligent computer system,a system that e&hibits the characteristics of intelligent in human beha iour,understanding language learning,recognise during problems. 'y 'ass 1 2degenbaum

3)AI is the study of how to make computer do things at which at present human beings is better. 4)AI is an engineering discipline is concerned with transforming the dum mode of functionality of a computer to made of operation which e&hibits more or less intelligence and are more natural for human being to cope with. Smart computer can help us in acti itie such as,# -))omputer as an informed office assistant. #produce a type script flow direction. #edit document from roughly prepared +specified instruction(spoken or written) #make abstact of documents #make type script of handwreitten documents. #retie e field information band a roughly specified re.uirement #plan tra el satisfying gi en constraints. /))omputer as knowledge teacher, #Take a sudent through a graded se.uence of lesson. #Interact with the student to clarify doubts,correct,misunderstanding etc. #Administrater test to e aluate understanding knowledge transfer ,grade a batch of student Authorware. 0))omputer as an 5&pert consultant #medical diagnosis#6%ycin #e.uipment maintenance and repair configuring e.uipments to create system with specified performance #Assistance in detailed design. #legal aderson 3))omputer as a free agent (robotics) #na igate +e&plore a terroism #manipulate ob!ects assemble e.uipments from kit.

Simu+ation o* !o-hi!ticated and Inte++i,ent beha.io in di**e ent a ea

The problem of simulating (or creating) intelligence has been broken down into a number of specific sub#problems. These consist of particular traits or capabilities that researchers would like an intelligent system to display. The traits described below ha e recei ed the most attention. 7eduction, reasoning, problem sol ing The different areas in which AI is applied are discussed below, Artificial Intelligence#(a)"eneral 8roblem Sol ing (b)5&pert System (c)$atural language processing (d))omputer ision (e)9obotics (f):thers (a)Gene a+ / ob+em So+.in,,It in ol es sol ing of broad range of problems which include reasoning about physical ob!ects and their relationship to each other as well as reasoning about actions and their conse.uences.Se eral problems like water !ug problem,tra ellinn salesman problemor other problems like tower of hanoi,%onkey and bananas problem or the missionaries and cannibles problem etc. (b)E0-e t !y!tem,These are AI programs which act as intelligent ad isor or consultant in a specific domain or specified areas.5&pert system do not replace e&pert but make a ailable their knowledge and e&perience.These systems are also a ailable for computer confi.uration and financial planning. (c)Natu a+ +an,ua,e - oce!!in,:$atural language means the nati e language i.e.the language one speaks.In order to understand natural language it must know how to ;"enerate ;<nderstand ;Translate %a!or components of natural language processing are, -)8arser /)=nowledge 9epresntation System 0):utput Translation 3)7ictionary (d)Com-ute .i!ion:It is the science and technology of machine that see.As a scientific discipline ,it is concerned with the theory and technology for building artificial

intelligence that obtain information fron images or multidimensional data.It is atechni.ue for a computer system to search beyond the data it is gi en and to find out almost the real world by analyzing and e aluating isual information. (e)Robotic!:9obotics is the field of engineering de oted to duplicating the physical capabilities of human being.attempt to mimic human mental abilities.They differ from AI programs which usually operate in computer simulated world whereas robots operate in physical world. (e)(the !:As AI field is de eloping it is e&pected that there are areas in which AI techni.ues of simulation of sophisticated and intelligent beha iour will be used in near future. Another field that is fast de eloping is speech recognition i.e.the computer>s ability to acceppt spoken words as dictation or to follow oice commands by using software.

/ ob+em !o+.in, in ,ame!: 1ATER 2&G /R(3LE$

(e are gi en / !ugs,a 3 gallononeand three gallon one.$either has any measuring markers on it.There is a pump which can be used to fill the !ug with water.?ow can we get e&actly / gallon of water into the 3 gallon !ug@ The state space for this problem can be described as the ordered set of pairs of integers(&,y)such that ABC,-,/,0 or 3 and DBC,-,/,0. A is the number of gallon of water in the 3 gallon !ug and D is the .uantity of water in the 0 gallon !ug. The start state for this problem is (C,C)and the goal state is (/,n) for any alue of n.There are -/ production rules for the water !ug problem

/ oduction Ru+e! Fo 1ate 2u, / ob+em

-.(A,D)if AE3 #6(3,D) #62ill the 3 gallon !ug. /. (A,D)ifDE0#6(A,0) #62ill the 0 gallon !ug. 0. (A,D)if A6C#6(A#d,D)#68our some water out of 3 gallon !ug. 3. (A,D)if D6C#6(A,D#d)#68our some water out of 0 gallon !ug. 4. (A,D)if A6C#6(C,D)#65mpty the 3 gallon !ug. F. (A,D)if D6C#6(A,C)#6 5mpty the 0 gallon !ug. G. (A,D)if AHD6B3 and D6C#6(3,D#(3#A))#68our water to 0 gallon !ug onto 3 gallon !ug. I. (A,D)if AHD6B0 and A6C#6(A#(0#D),0))#68our water to 3 gallon !ug onto 0 gallon!ug. J. (A,D)if AHDEB3 and D6C#6(AHD,C)#68our all the water from 0 gallon !ug into 3 gallon !ug.

-C. (A,D) if AHDEB0 and A6C#6(C,AHD)#68our all the water from 3 gallon !ug into 0 gallon !ug. --. (C,/)#6(/,C)#68our the / gallon water from 0 gallon !ug into 3 gallon !ug. -/. (/,D)#6(C,D)#65mpty the / gallon in the 3 gallon !ug.

Natu a+ +an,ua,e :
$atural language processing,$atural language processing gi es machines the ability to read and understand the languages that the human beings speak. %any researchers hope that a sufficiently powerful natural language processing system would be able to ac.uire knowledge on its own, by reading the e&isting te&t a ailable o er the internet. Some straightforward applications of natural language processing include information retrie al (or te&t mining) and machine translation

Automated ea!onin,:
Automated reasoning is an area of computer science dedicated to understanding different aspects of reasoning in a way that allows the creation of software which allows computers to reason completely or nearly completely automatically. As such, it is usually considered a subfield of artificial intelligence, but it also has strong connections to theoretical computer science and e en philosophy. The most de eloped sub areas of automated reasoning probably are automated theorem pro ing (and the less automated but more pragmatic subfield of interacti e theorem pro ing) and automated proof checking ( iewed as guaranteed correct reasoning under fi&ed assumptions), but e&tensi e work has also been done in reasoning by analogy induction and abduction. :ther important topics are reasoning under uncertainty and non# monotonic reasoning. An important part of the uncertainty field is that of argumentation, where further constraints of minimalist and consistency are applied on top of the more standard automated deduction. Kohn 8ollockLs :scar system is an e&ample of an automated argumentation system that is more specific than being !ust an automated theorem pro erb. Tools and techni.ues include the classical logics and calculi from automated theorem pro ing, but also fuzzy logic, 'ayesian inference, reasoning with ma&imal entropy and a large number of less formal ad#hoc techni.ues.

4i!ua+ -e ce-tion :
Misual perception is the ability to interpret information from isible light reaching the eye. The resulting perception is also known as eyesight, sight or ision. The arious physiological components in ol ed in ision are referred to collecti ely as the isual system, and are the focus of much research in psychology, cogniti e science, 4i!ua+ !y!tem: The isual system in humans allows indi iduals to assimilate information from the en ironment. The act of seeing starts when the lens of the eye focuses an image of its surroundings onto a light#sensiti e membrane in the back of the eye, called the retina.

The retina is actually part of the brain that is isolated to ser e as a transducer for the con ersion of patterns of light into neuronal signals. The lens of the eye focuses light on the photorecepti e cells of the retina, which detect the photons of light and respond by producing neural impulses. These signals are processed in a hierarchical fashion by different parts of the brain, from the retina to the lateral geniculate nucleus, to the primary and secondary isual corte& of the brain.

5eu i!tic a+,o ithm:

In computer science, a heuristic algorithm, or simply a heuristic, is an algorithm that is able to produce an acceptable solution to a problem in many practical scenarios, but for which there is no formal proof of its correctness. Alternati ely, it may be correct, but may not be pro en to produce an optimal solution, or to use reasonable resources. ?euristics are typically used when there is no known method to find an optimal solution. Two fundamental goals in computer science are finding algorithms with pro ably good run times and with pro ably good or optimal solution .uality. A heuristic is an algorithm that abandons one or both of these goalsN for e&ample, it usually finds pretty good solutions, but there is no proof the solutions could not get arbitrarily badN or it usually runs reasonably .uickly, but there is no argument that this will always be the case. ?euristic search algorithm are, -)"enerate and test /)?illclimbing 0)'est first search 3)'ranch and 'ound 4)8roblem 9eduction F))onstraint Satisfying G)%eans end analysis. A+,o ithm o* 3e!t *i !t !ea ch: -)8ut the initial node on a list :85$. /)If(:85$ is empty) or (:85$ B":AO)terminate search. 0)9emo e the first node from :85$.)all this node a. 3)If (aB":AO)terminate search with success. 4)5lse if node a has successors, generate all of them.2ind out how far they are from the goal node.Sort all the children generated so far by the remaining distance from the goal. F)$ame this list as )O:S57. G)9eplace :85$ with )O:S57. I)"o to step /.

A+,o ithm *o A6 -)8ut the initial node on alist :85$. /)If(:85$ is empty)or (:85$ B":AO)terminate search. 0)9emo e the first node from :85$.)all this node a. 3)If (aB":AO)terminate search with success. 4)5lse if node a has successors,generate all of them.5stimate the fitness number of the successors by totalling the e aluation function alue and the cost#function alue.Sort the list by fitness number. F)$ame the new list as)O:S57. G)9eplace :85$ with )O:S57. I)"oto step /.

So+ution ,ua anteed a+,o ithm:

)e-th-*i !t !ea ch: 7epth#first search is an algorithm for tra ersing or searching a tree, tree structure, or graph. :ne starts at the root (selecting some node as the root in the graph case) and e&plores as far as possible along each branch before backtracking. A+,o ithm o* )FS: -)if n is a goal node then, solution BnN /)for each successor n i of n do if n i is an ancestor of n then <72S(n i)N )e*inition: 2ormally 72S is an uninformed search that progresses by e&panding the first child node of the search tree that appears and thus going deeper and deeper until a goal node is found or until it hits a node that has no children. Then the search backtracks returning to the most recent node it hasnLt finished e&ploring. In a non#recursi e implementation, all freshly e&panded nodes are added to a stack for e&ploration. Space comple&ity of 72S is much lower than '2S (breadth#first search). It also lends itself much better to heuristic methods of choosing a likely#looking branch. Time comple&ity of both algorithms is proportional to the number of ertices plus the number of edges in the graphs they tra erse (hen searching large graphs that cannot be fully contained in memory, 72S suffers from non#termination when the length of a path in the search tree is infinite.

3 eadth-*i !t !ea ch: In graph theory, breadth#first search is a graph search algorithm that begins at the root node and e&plores all the neighboring nodes. Then for each of those nearest nodes, it e&plores their une&plored neighbor nodes, and so on, until it finds the goal. '2S is an uninformed search method that aims to e&pand and e&amine all nodes of a graph or combination of se.uences by systematically searching through e ery solution. In other words, it e&hausti ely searches the entire graph or se.uence without considering the goal until it finds it. It does not use a heuristic algorithm. A+,o ithm o* 3FS, -.8ut the start node on a list,called :85$ of une&plored nodes.If the start node is goal node,a solution has been found. /)If :85$ is empty ,no solution e&ist. 0)9emo e the first node,n from :85$ and place it in a list called )O:S57,of e&panded nodes. 3)5&pand node n.If it has no successor go to (/). 4) place all successors of node n at the end of:85$. F)If any of the successors of node n is a goal state,solution has been found.

&NIT7: &N)ERSTAN)ING NAT&RAL LANG&AGE /a !in, techni8ue!:

In computer science and linguistics, parsing, or, more formally, syntactic analysis, is the process of analyzing a se.uence of tokens (for e&ample, words) to determine their grammatical structure with respect to a gi en (more or less) formal grammar. 8arsing is also an earlier term for the diagramming of sentences of natural languages, and is still used for the diagramming of inflected languages. S#6$8M8 $8#6the $8 $8#689:

$8#6$8 $8#6A7KS $ ADJS->A7K A7KS M8#6M M8#6M $8 $#62ile printer 8$#6San!ay 89:#6 A7K#6shortlongfast M#6printedcreatedwant A simple grammar for a fragment of 5nglish.

Ty-e! o* -a !e ! The task of the parser is essentially to determine if and how the input can be deri ed from the start symbol of the grammar. This can be done in essentially two ways, '%To--do9n -a !in, # Top#down parsing can be iewed as an attempt to find left# most deri ations of an input#stream by searching for parse trees using a top#down e&pansion of the gi en formal grammar rules. Tokens are consumed from left to right. Inclusi e choice is used to accommodate ambiguity by e&panding all alternati e right#hand#sides of grammar rules OO parsers and recursi e#descent parser are e&amples of top#down parsers which cannot accommodate left recursi e productions. Although it has been belie ed that simple implementations of top#down parsing cannot accommodate direct and indirect left#recursion and may re.uire e&ponential time and space comple&ity while parsing ambiguous conte&t#free grammars, more sophisticated algorithms for top#down parsing ha e been created by 2rost, ?afiz, and )allaghan which accommodate ambiguity and left recursion in polynomial time and which generates polynomial#size representations of the potentially#e&ponential number of parse trees. Their algorithm is able to produce both left#most and right# most deri ations of an input with regard to a gi en )2". E0am-+e! o* To--do9n -a !e ! Some of the parsers that use top#down parsing include, -)9ecursi e descent parser /)OO parser (Oeft#to#right, Oeftmost deri ation)

7%3ottom-u- -a !in, # A parser can start with the input and attempt to rewrite it to the start symbol. Intuiti ely, the parser attempts to locate the most basic elements, then the elements containing these, and so on. O9 parsers are e&amples of bottom#up parsers. Another term used for this type of parser is Shift#9educe parsing. . E0am-+e! o* 3ottom-u- -a !e ! Some of the parsers that use bottom#up parsing include, -)8recedence parser -):perator#precedence parser /)Simple precedence parser /)') (bounded conte&t) parsing 0)O9 parser (Oeft#to#right, 9ightmost deri ation) -)Simple O9 (SO9) parser /)OAO9 parser 0))anonical O9 (O9(-)) parser

Context-free grammar:
In formal language theory, a conte&t#free grammar ()2") is a grammar in which e ery production rule is of the form MPw where M is a single nonterminal symbol, and w is a string of terminals and nonterminals Thus, the difference with arbitrary grammars is that the left hand side of a production rule is always a single nonterminal symbol rather than a string of terminal or nonterminal symbols. The term conte&t#free e&presses the fact that nonterminals are rewritten without regard to the conte&t in which they occur. A formal language is conte&t#free if some conte&t#free grammar generates it. These languages are e&actly all languages that can be recognized by a non#deterministic pushdown automaton. )onte&t#free grammars play a central role in the description and design of programming languages and compilers. They are also used for analyzing the synta& of natural languages

Fo ma+ de*inition!
A conte&t#free grammar G is a 3#tuple, where -. A finite set of non#terminal characters or ariables. They represent different types of phrase or clause in the sentence.

/. A finite set of terminals, dis!oint with , which make up the actual content of the sentence. 0. A relation from to such that the symbol belongs to the set. . 3. The start ariable, used to represent the whole sentence (or program). It must be an element of set. The members are called the rules or productions of the grammar.

T an!*o mationa+ , amma :

In linguistics, a transformational grammar, or transformational#generati e grammar (T""), is a generati e grammar, especially of a natural language, that has been de eloped in a )homskyan tradition. Additionally, transformational grammar is the )homskyan tradition that gi es rise to specific transformational grammars. %uch current research in transformational grammar is inspired by )homskyLs %inimalist 8rogram.

T an!*o mation!
The usual usage of the term LtransformationL in linguistics refers to a rule that takes an input typically called the 7eep Structure or 7#structure and changes it in some restricted way to result in a Surface Structure.or s structure.In T"" 7eep structures were generated by a set of phrase structure rules. Fo e0am-+e a typical transformation in T" is the operation of sub!ect#au&iliary in ersion .This rule takes as its input a declarati e sentence with an au&iliary, Kohn has eaten all the heirloom tomatoes. and transforms it into ?as Kohn eaten all the heirloom tomatoes. In their original formulation , these rules were stated as rules that held o er strings of either terminals or constituent symbols or both. A $8 A<A D A A<A $8 D (where $8 B $oun 8hrase and A<A B Au&iliary)

T an!ition net!:
A Transition net consists of the following constituents, -. A directed net (S, T, 2). /. A structure consisting of some sorts of indi iduals together with some operations and relations. 0. A labeling of arcs assigning to all elements of 2 a formal sum of n#tuples of ariables where n is the arity of the predicate connected to the arc 3. An inscription on transitions assigning to some elements of T a logical formula built from Q. where ariables occurring free in a transition ha e to occur at an ad!acent arc

4. A marking % of predicates S with formal sums of n#tuples of indi iduals. F. A function = which assigns to the predicates an upper bound for the number of copies of the same item which it may carry. G. The transition rule which e&presses the common interpretation of 8rt+T nets, A transition may occur, for an assignment of indi iduals to the ariables which satisfy the formula inscribed to the transition, if all input predicates carry enough copies of proper items and for no output predicate the capacity = is e&ceeded by adding the respecti e copies of items. The set of all markings connected to % through such occurrences of transitions is denoted by R%6S.

Au,mented t an!ition net!:

Augmented place transition nets with token alues are used as a semantic data modeling formalism. This formalism is able to co er rele ant aspects of the dynamic beha ior of a database in a concise manner and enables the data engineer to specify constraints for e ent se.uences and ob!ect type assignmentsN therefore it is proposed as a data modeling tool to map ob!ect#e ent interactions from the uni erse of discourse to the conceptual scheme.

Fi++mo e:! , amma :

According to 2illmore, each erb selects a certain number of deep cases which form its case frame. Thus, a case frame describes important aspects of semantic alency, of erbs, ad!ecti es and nouns. )ase frames are sub!ect to certain constraints, such as that a deep case can occur only once per sentence. Some of the cases are obligatory and others are optional. :bligatory cases may not be deleted, at the risk of producing ungrammatical sentences. 2or e&ample, %ary ga e the apples is ungrammatical in this sense. A fundamental hypothesis of case grammar is that grammatical functions, such as sub!ect or ob!ect, are determined by the deep, semantic alence of the erb, which finds its syntactic correlate in such grammatical categories as Sub!ect and :b!ect, and in grammatical cases such as $ominati e, Accusati e, etc. 2illmore puts forwards the following hierarchy for a uni ersal sub!ect selection rule, Agent E Instrumental E :b!ecti e That means that if the case frame of a erb contains an agent, this one is realized as the sub!ect of an acti e sentence, otherwise the deep case following the agent in the hierarchy is promoted to sub!ect.

Shan"! conce-tua+ de-endency:

$O< Systems 8arsing is a method of decomposing or breaking apart 5nglish statements in an attempt to better understand relationships within the te&t. These techni.ues work from the phonological le el up through the syntactic and e en semantic le els. 8arse trees # which are similar to sentence diagrams are used to classify words and clauses in a sentence. Some ariations on parsing include transitional grammars, which focus on inter and intra sentence transition, nondeterministic methods, which can use bottom up or top down analysis, deterministic methods, which delay analysis until a sufficient look ahead is complete, and augmented transition nets which force sentences into pre#arranged

relationships. )onceptual 7ependency 7iagramming is a method which operates primarily at the semantic le el. 9oger Shank, while at Dale de eloped his own calculus of ob!ects, relationships, and symbolic manipulation. )77 could be iewed as e&tensions to AT$s. The thrust of ShankLs system was to place all language within a finite logical set of ob!ects or relationships. In this manner the underlying TmeaningT of the te&t could be understood. :nce established, a )77 could be mo ed into a frame or script to deri e e en further general knowledge about a situation. Another ery recent effort in building theory in the $O< arena has been the %essage <nderstanding )onferences. 2our classes of concepts, picture producers (88), actions(A)T), picture aiders(8A),and action aiders (AA) which ery loosely correspond to the syntactic categories of nouns, erbs, ad!ecti es, and ad erbs respecti ely. 8resents a formalism called )#diagrams which represent relationships between concepts and conceptualizations. It describes the processing which goes into forming such representations including resolution of issues such as meaningfulness constraints, syntactic ambiguity, semantic ambiguity, etc.

G amma * ee ana+y;e :
)onte&t#2ree "rammar, A conte&t#free grammar " is a .uadruple (M, , 9, S), where M is an alphabet, (the set of terminals) is a subset of M, 9 (the set of rules) is a finite subset of (M#) M;, and S (the start symbol) is an element of M#. $ot all conte&t# free languages are regular. 5 ery regular language is conte&t# free.

Sentence ,ene ation

Sentence generation is the computational process of automatically producing sentences in Some human language on the basis of a specification of communicati e intention. A sentence generation component recei es as input a specification of what it is supposed to communicate, and produces as output a corresponding natural language e&pression. In most current approaches to sentence generation, such input is either a logical form or a set of facts in some knowledge representation language. Thus, a generation component might accept as input a logical form such as, :n(&, y). 'ookU(&).redU(&). tableU(y) And con ert this into something like the natural language e&pression. The red book is on the table. The end result of such a generation process is relati ely clear .A sentence must be produced which is grammatically correct and which has made appropriate choices of words (or le&emes).

T an!+ation
Translation memories tools often pro ide automatic retrie al and substitution. Automatic et ie.a+

T%s are searched and displayed automatically as a translator mo es through a document. Automatic !ub!titution 5&act matches come up in translating new ersions of a document. 7uring automatic substitution, the translator does not check the translation against the original, so if there are any mistakes in the pre ious translation, they will carry o er.

&NIT 3 <N(1LE)GE RE/RESENTATI(N:

Fi !t o de - edicate ca+cu+u!
2irst#:rder Oogic 9epresentation %any of the architectures analyzed build upon a substrate of 2irst :rder 8redicate )alculus. This is a ery descripti e declarati e representation with a well founded method of deri ing new knowledge from a database. Its fle&ibility makes it a good choice when more than one module may add to or utilize a common database. <nfortunately is a fle&ibility limitation. To maintain consistency, learning must be monotonic. This limits its effecti eness when there are incomplete domain theories.

2irst#order predicate logic is composed of statements that are assumed to be true. The statements are composed of, ; Atoms (symbols), ; 8redicates (a function with one or more atomic arguments), ; Two sub statements !oined by a con!unction, dis!unction, or implication, ; A negated sub statement, and ; A statement with an e&istential or uni ersal .uantifier. Synta0 *o F(/L: The basic syntactic elements of first order logic are the symbols which stand for ob!ect, relations, and functions. These are of three kinds, Con!tant !ymbo+! which stand for ob!ects. / edicate !ymbo+! which stands for relation. Function !ymbo+ which stands for functions. The symbols and rules of 2:8O are, -) )onstants, )onstant are fi&ed alue terms that belong to a gi en domain of discourse. /) 8redicate, 8redicate symbols denote relations or functional mapping of element from the domain. 0) 2unctions, 2unctions symbols denote relation defined on a domain. 3) Mariables, Mariables are terms that can assume different alues o er a gi en domain. 4) Vuantifiers, The two .uantifier symbols are a) 5&istential .uantifier b) <ni ersal .uantifier F) )onnecti es, 2:8O uses 4 connecti es i.e. not, or, and, implication and e.ui alence

5o n c+au!e
In mathematical logic, a ?orn clause is a clause (a dis!unction of literals) with at most one positi e literal. A ?orn clause with e&actly one positi e literal is a definite clause. A horn clause with no positi e literals is called goal clause. in uni ersal algebra definite. A ?orn clause with no positi e literals is sometimes called a goal clause or .uery clause, especially in logic programming. A dual#?orn clause is a clause with at most one negati e literal. ?orn clauses play a basic role in logic programming and are important for constructi e logic. ?orn clauses can be propositional or first order, depending on whether we consider propositional or first#order literals. The rele ance of ?orn clauses to theorem pro ing by first#order resolution is that the resolution of two ?orn clauses is a ?orn clause. %oreo er, the resolution of a goal clause and a definite clause is again a goal

In logic programming a definite clause beha es as a goal#reduction procedure. 8ropositional ?orn clauses are also of interest in computational comple&ity, where the problem of finding a set of ariable assignments to make a con!unction of ?orn clauses true is a 8#complete problem, sometimes called ?:9$SAT. This is 8Ls ersion of the 'oolean satisfiability problem, a central $8#complete problem. Satisfiability of first# order ?orn clauses is undecidable.

/ o+o,
8rolog is a logic programming general purpose fifth generation language associated with artificial intelligence and computational linguistics. It has a purely logical subset, called pure 8rolog, as well as a number of e&tra logical features.8rolog has its roots in formal logic, and unlike many other programming languages, 8rolog is declarati e, The program logic is e&pressed in terms of relations, and e&ecution is triggered by running .ueries o er these relations. 9elations and .ueries are constructed using 8rologLs single data type, the term. 9elations are defined by clauses 8rolog was one of the first logic programming languages, and remains among the most popular such languages today, with many free and commercial implementations a ailable. (hile initially aimed at natural language processing, the language has since then stretched far into other areas like theorem pro ing, e&pert systems, games, automated answering systems, ontologies and sophisticated control systems, and modern 8rolog en ironments support the creation of graphical user interfaces, as well as administrati e and networked applications. Subse.uent e&tensions of 8rolog by the

Semantic net!
A semantic network is a network which represents semantic relations between the concepts. This is often used as a form of knowledge representation. It is a directed or undirected graph consisting of ertices, which represent concepts, and edges An e&ample of a semantic network is (ord $et, a le&ical database of 5nglish. It groups 5nglish words into sets of synonyms called synsets, pro ides short, general definitions, and records the arious semantic relations between these synonym sets. Some of the most common semantic relations defined are metonymy (A is part of ', i.e. ' has A as a part of itself), homonymy (' is part of A, i.e. A has ' as a part of itself), hyponymy (or troponymy) (A is subordinate of 'N A is kind of '), hypernymy (A is super ordinate of '), synonymy (A denotes the same as ') and antonyms (A denotes the opposite of '). Ty-e! o* !emantic net! -) 7efinitional network, It emphasizes the subtype or is#a relation between a concept type and a newly defined subtype. The resulting network also called a generalization or subsumption hierarchy. /) Assert ional network, Are designed to assert propositions. Some assert ional networks ha e been proposed as models of conceptual structures underlying natural language semantic. 0) Implication networks, use implication as the primary relation for connecting

nodes. 3) 5&ecutable networks, include some mechanism, such as marker passing or attached procedures, which can perform inferences, pass messages, or search for pattern and associations. 4) Oearning network build or e&tend their representation by ac.uiring knowledge from e&amples. The new knowledge may change the old network by adding and deleting nodes and arcs or by modifying the numerical alues called weight. F) ?ybrid $etwork, combine two or more of the pre ious techni.ues, either in a single network or in separate.

/a titioned Net9o "!

8artitioned Semantic $etworks allow for, ; 8ropositions to be made without commitment to truth. ; 5&pressions to be .uantified. 'reak network into spaces which consist of groups of nodes and arcs and regard each space as a node. 5&ample, 5 ery parent lo es their child To represent this we, ; )reate a general statement, "S, special class. ; %ake node g an instance of "S. ; 5 ery element will ha e at least / attributes, o a form that states which relation is being asserted. o one or more for all () or e&ists () connections # these represent uni ersally .uantifiable ariables in such statements e.g. in parent(&), child(y) lo es(&,y)

$in!"ey F ame
(hen one encounters a new situation, one selects from memory a structure called frame. A frame is a data structure for representing a stereotyped situation. )ollections of related frames are linked together into frame#system. In the field of Artificial Intelligence, a frame is a data structure introduced by %ar in %insky in the -JGCs that can be used for knowledge representation. %insky frames are intended to help an Artificial Intelligence system recognize specific instances of patterns. 2rames usually contain properties called attributes or slots. Slots may contain default alues, refer to other frames or contain methods for recognizing pattern instances. 2rames are thus a machine#usable formalization of concepts or schemata. In contrast, the ob!ect#oriented paradigm partitions an information domain into abstraction hierarchies rather than partitioning into component

hierarchies, and is used to implement any kind of information processing. 2rame Technology is loosely based on %insky 2rames, its purpose being software synthesis rather than pattern analysis. Oike many other knowledge representation systems and languages, frames are an attempt to resemble the way human beings are storing knowledge.. In frame#based knowledge representations knowledge describing a particular concept is organized as a frame. The frame usually contains a name and a set of slots.

Ca!e , amma theo y

)ase "rammar is a system of linguistic analysis, focusing on the link between the alence of a erb and the grammatical conte&t it re.uires, created by the American linguist )harles K. 2illmore in (-JFI), in the conte&t of Transformational "rammar. This theory analyzes the surface syntactic structure of sentences by studying the combination of deep cases (i.e. semantic roles) Agent, :b!ect, 'enefactor, Oocation or Instrument which are re.uired by a specific erb. The following list cases are, a) Source(S) b) Oocati e (O) c) Agent (A) d) 2actiti e (2) e) :b!ect (:) f) Instrument (I) g) Time (T) h) "oal (") i) 7irection (7) 2or instance, the erb gi e in 5nglish re.uires an Agent (A) and :b!ect (:), and a 'eneficiary (')N e.g. TKones (A) ga e money (:) to the school ('). According to 2illmore, each erb selects a certain number of deep cases which form its case frame. Thus, a case frame describes important aspects of semantic alences, of erbs, ad!ecti es and nouns. )ase frames are sub!ect to certain constraints, such as that a deep case can occur only once per sentence. Some of the cases are obligatory and others are optional. :bligatory cases may not be deleted, at the risk of producing ungrammatical sentences. 2or e&ample, %ary ga e the apples is ungrammatical in this sense. 2illmore (-JFI) puts forwards the following hierarchy for a uni ersal sub!ect grammar is that grammatical functions, such as sub!ect or ob!ect, are determined by the deep, semantic alence of the erb, which finds its syntactic correlate in such grammatical categories selection rule, Agent E Instrumental E :b!ecti e That means that if the case frame of a erb contains an agent, this one is realized as the sub!ect of an acti e sentenceN otherwise, the deep case following the agent in the hierarchy is promoted to sub!ect.

/ oduction Ru+e! *o <no9+ed,e Re- e!entation

:ne of the most popular approaches to knowledge representation is to use production rules, sometimes called I2#T?5$ rules. They can take arious forms, e.g I2 condition T?5$ action I2 premise T?5$ conclusion I2 proposition p- and proposition p/ are true T?5$ proposition p0 is true Some of the benefits of I2#T?5$ rules are that they are modular, each defining a relati ely small and, at least in principle, independent piece of knowledge. $ew rules may be added and old ones deleted usually independently of other rules. %ycin was designed to help the doctor to decide whether a patient has a bacterial infection, which organism is responsible, which drug may be appropriate for this infection, and which may be used on the specific patient. The global knowledge base contains facts and rules relating for e&ample symptoms to infections, and the local database will contain particular obser ations about the patient being e&amined. A typical rule in %ycin is as follows, I2 the identity of the germ is not known with certainty A$7 the germ is gram#positi e A$7 the morphology of the organism is rod A$7 the germ is aerobic T?5$ there is a strong probability

The Inte *ace Sy!tem

5 en the most sophisticated e&pert system is worthless if the intended user cannot communicate with it. The components of an e&pert system which communicate with the user is known as interface system or user interface. The communication performed by a user interface is bidirectional. %ost user interfaces make hea y use of techni.ues de eloped in another artificial intelligence discipline, natural language processing. These techni.ues show us how to communicate with an e&pert system in ordinary 5nglish and enables the computer to respond to us in the same language.

INFERENCE S=STE$
Inference is the act or process of deri ing a logical conse.uence conclusion from premises. Inference is studied within se eral different fields. ; ?uman inference (i.e. how humans draw conclusions) is traditionally studied within the field of cogniti e psychology.

; Oogic studies the laws of alid inference. ; Statisticians ha e de eloped formal rules for inference (statistical inference) from .uantitati e data. ; Artificial intelligence researchers de elop automated inference systems.

T=/ES (F INFERENCES 6Inducti.e

The process by which a conclusion is inferred from multiple obser ations is called inducti e reasoning. The conclusion may be correct or incorrect, or correct, or correct to within a certain degree of accuracy, or correct in certain situations. )onclusions inferred from multiple obser ations may be tested by additional obser ations. 6)educti.e The process by which a conclusion is logically inferred from certain premises is called deducti e reasoning. %athematics makes use of deducti e inference. 64a+id in*e ence! Inferences are either alid or in alid, but not both. 8hilosophical logic has attempted to define the rules of proper inference, i.e. the formal rules that, when correctly applied to true premises, lead to true conclusions.

Fo 9a d and 3ac"9a d deduction:

There are two important ways in which rules can be used in a rule based system to draw inferences. -) 2orward chaining /) 'ackward chaining In 2orward )haining the search for new information seems to be proceeding in the direction of the arrow separating the left hand side and right hand sides of the rules. 'ackward )haining is more cost effecti e.

&NIT>: E?/ERT S=STE$

Int oduction:
An e&pert system consists of the following main components, -) =nowledge base /) Inference 5ngine#a reasoning mechanism and heuristics for problem sol ing (search techni.ues) 0) 5&planatory components 3) ?uman#machine interface Featu e! o* an E0-e t Sy!tem The general features of an e&pert system are, -) ?igh le el e&pertise, The most useful features of an e&pert system is the high le el e&pertise it pro ides to aid in problem sol ing. /) 8redicti e modeling power, Another useful feature of an e&pert system is its predicti e modeling power. The system can acts as information processing theory or model of problem sol ing in the gi en domain. 0) Institutional memory, The corpus of knowledge which defines the proficiency of an e&pert system can also pro ide an additional feature, an institutional memory. 3) Ability to pro ide Training facility , A final features of an e&pert system is its ability to pro ide a training facility for key personnel and important members. E0-e t Sy!tem A chitectu e:

There are generally two types of architectures that are used in e&pert systems. They are, -) 9ule based architectures /) $on production system architectures Ru+e ba!ed a chitectu e!: This type of system uses knowledge encoded in the form of production rules, i.e. I2#T?5$ rules. 5ach rule represents a small chunk of knowledge relating to the gi en domain of e&pertise. The main components of a typical e&pert system are, a) =nowledge 'ase b) The Inference 8rocess c) The Input W:utput interface. Non- oduction Sy!tem A chitectu e!: Instead of rules, these systems employ more structured representation schemes like associati e or semantic networks, frame and rule structures and neural network. The no production system architectures are, -) Associati e or semantic network architectures /) 2rame architectures 0) 7ecision Tree architectures 3) Analogical reasoning architectures 4) $eural network architectures

E0i!tin, E0-e t Sy!tem )end a+

7endral was an influential pioneer pro!ect in artificial intelligence (AI) of the -JFCs, and the computer software e&pert system that it produced. Its primary aim was to help organic chemists in identifying unknown organic molecules, by analyzing their mass spectra and using knowledge of chemistry. It was done at Stanford <ni ersity by 5dward 2eigenbaum. The software program 7endral is considered the first e&pert system because it automated the decision#making process and problem#sol ing beha ior of organic chemists. It consists of two sub#programs, ?euristic 7endral and %eta#7endral. It was written in Oisp (programming language). %any systems were deri ed from 7endral, including %D)I$, %:O"5$, %A)SD%A, 89:S85)T:9, A):$, and ST5A%59. 5eu i!tic )end a+ ?euristic 7endral is a program that uses mass spectra or other e&perimental data together with knowledge base of chemistry, to produce a set of possible chemical structures that

may be responsible for producing the data. %ass spectrum of a compound is produced by a mass spectrometer, and is used to determine its molecular weight, the sum of the masses of its atomic constituents. 2or e&ample, the compound water (?/:), has a molecular weight of -I since hydrogen has a mass of -.C- and o&ygen -F.CC, and its mass spectrum has a peak at -I units. ?euristic 7endral would use this input mass and the knowledge of atomic mass numbers and alence rules, to determine the possible combinations of atomic constituents whose mass would add up to -I.As the weight increases and the molecules become more comple&, the number of possible compounds increases drastically. Thus, a program that is able to reduce this number of candidate solutions through the process of hypothesis formation is essential. $eta-)end a+ %eta#7endral is a knowledge ac.uisition system that recei es the set of possible chemical structures and corresponding mass spectra as input, and proposes a set of hypotheses to e&plain correlation between some of the proposed structures and the mass spectrum. These hypotheses would be fed back to ?euristic 7endral to test their applicability. Thus, T?euristic 7endral is a performance system and %eta#7endral is a learning systemT.The program is based on two important features, the plan#generate#test paradigm and knowledge engineering.

$ycin
%D)I$ was an early e&pert system de eloped o er fi e or si& years in the early -JGCs at Stanford <ni ersity. It was written in Oisp as the doctoral dissertation of 5dward Shortleaf under the direction of 'ruce 'uchanan, Stanley $. )ohen and others. It arose in the laboratory that had created the earlier 7endral e&pert system, but emphasized the use of !udgmental rules that had elements of uncertainty associated with them. This e&pert system was designed to identify bacteria causing se ere infections, such as bacterium and meningitis, and to recommend antibiotics, with the dosage ad!usted for patientLs body weight #the name deri ed from the antibiotics themsel es, as many antibiotics ha e the suffi& Wmycin.

A--+ication o* E0-e t Sy!tem Techno+o,y

a) )ontrolling and monitoring b) 7ebugging c) 7esign d) 7iagnosis e) Instruction f) Interpretation g) 8lanning h) 8rediction

)omain E0-+o ation:

The domain refers to the knowledge that is part of the world model the system knows about. This includes ob!ect description, relationships and other concepts. 7omain knowledge is the knowledge which is alid and directly used for a reselected domain of human or an autonomous computer acti ity. In software engineering, domain knowledge is that knowledge which software programs encode in the form of different knowledge bases. :perations on domain knowledge are performed by meta knowledge. 5&amples, -) (ord processors encode specific knowledge about fonts and formatting of te&t. /) 5mbedded controllers for automobiles engines know how to control engines. 0) %edical software encodes knowledge and information about patience and medicine. 3) 2inancial software encodes formal rules related to purchasing, Accounts, and in entory.

$eta"no9+ed,e
%etaknowledge or meta#knowledge is knowledge about a preselected knowledge. 2or the reason of different definitions of knowledge in the sub!ect matter literature, meta# information is or is not included in meta#knowledge. 7etailed cogniti e, systemic and epistemic study of human knowledge re.uires a distinguishing of these concepts. 'ut in the common language knowledge includes information, and, for e&ample, bibliographic data are considered as a meta#knowledge. %eta#knowledge is a fundamental conceptual instrument in such research and scientific domains as, knowledge engineering, knowledge management, and others dealing with study and operations on knowledge, seen as unified ob!ect+entities, abstracted from local conceptualizations and terminologies. 5&amples of the first#le el indi idual meta# knowledge are methods of planning, modeling, learning and e ery modification of domain knowledge. <ni ersal meta#knowledge frameworks ha e to be alid for the organization of meta#le els of indi idual meta#knowledge.

Se+*-E0-+ainin, !y!tem
%ost of the e&pert system ha e self e&plaining facility,i.e why it asked certain .uestion, how it arri ed at its answers etc. To respond to how .uery, the e&planation module traces the chain of rules fired during a consultation with the user. If the user does not agree with the reasoning steps presented, they may be changed using the editor. To respond to why .uery, the e&planation module must be able to e&plain why certain information is needed by the inference engine to complete a step in reasoning process before it can proceed. ?owe er, selfe&plaining systems are not often used in an e&pert system.

$ajo / ob+em! Facin, cu ent E0-e t Sy!tem: -) 'rittleness, As e&pert system only ha e access to highly specific domain knowledge, they cannot fallback on more general knowledge when the need arises. /) Oack of metaknowledge, 5&pert system do not ha e ery sophisticatedd knowledge about their own operation. 0) =nowledge ac.uisition, 7espite the de elopment of arious tools for knowledge ac.uisition, it still remains a ma!or bottleneck in applying e&pert systems technology to new domain. 3) Malidation, %easuring the performance of an e&pert system is difficult because we do not know how to .uantify the use of knowledge.

&nit @ Int oduction to /atte n Reco,nition:

9ecognition is the process of established a close match between some new stimulus and pre iously stored stimulus patterns. This process is being performed continually through the li es of all li ing things. In higher animals this ability is manifested in many forms at both the conscious and unconscious le els for both abstract as well as physical ob!ects. 8attern recognition is a sub#topic of machine learning. It is the act of taking in raw data and taking an action based on the category of the data. %ost research in pattern recognition is about methods for super ised learning and unsuper ised learning. 8attern recognition aims to classify data (patterns) based either on a priori knowledge or on statistical information e&tracted from the patterns. The patterns to be classified are usually groups of measurements or obser ations, defining points in an appropriate multidimensional space. This is in contrast to pattern matching, where the pattern is rigidly specified.

The eco,nition and C+a!!i*ication / oce!!:

In artificial or mechanical recognition the steps performed are as follows, -. Sensors /. 2eature selection 0. %atching 3. )lassification rules

A-- oache! *o Reco,nition:

There are two basic approaches to recognition are as follows, -) Structured description /) Statistical description

') St uctu ed de!c i-tion: Syntactical (or structural) pattern recognition is based on the structural interrelationships of features.

7% Stati!tica+ de!c i-tion:

, Statistical pattern recognition is based on statistical characterizations of patterns, assuming that the patterns are generated by a probabilistic system.

St uctu ed )e!c i-tion:

Syntactic pattern recognition or structural pattern recognition is a form of pattern recognition, where items are presented in pattern structured which can take into account more comple& interrelationship between features. In statistical classification simple numerical features ectors are used. Structure 7escription is a family of knowledge representation languages which can be used to represent the concept definitions of an application domain in a structured and formally well#understood way. The name structure description logic refers, on the one hand, to concept descriptions used to describe a domain and, on the other hand, to the logic#based semantics which can be gi en by a translation into first#order predicate logic. Structure 7escription logic was designed as an e&tension to frames and semantic networks, which were not e.uipped with formal logic#based semantics. They form a middle ground solution, including some more e&pressi e operations than propositional logic and ha ing decidable or more efficient decision problems than first order predicate logic.

Symbo+ic )e!c i-tion:

It is important to distinguish between the mechanism for building the meaning of the fake word+ phrase and the mechanism that uses that meaning to build a .uestion with the word+ phrase as an answer. Symbolic description is di ided into three parts are as follows, -) Le0icon, There is a set of le&emes. A le&eme is an abstract entity roughly corresponding to a meaning of a word or phrase. 5ach le&eme has e&actly on entry in the le&icon, so if a word has two meanings it will ha e two corresponding le&emes /) Schemata: The word schema means a large set of facts with ariables. A schema or symbolic description stipulates asset of relationships which must hold between the le&emes used to build for a e&ample, a !oke 0) Tem-+ate!: A templates is used to produce the surface from of a !oke from the le&emes and relationships specified in an instantiated schema. Templates are not inherently humor#related gi en a noun phrase, and a meaning for that noun phrase a, templates builds a suitable .uestion answer pair.

$achine /e ce-tion:
8erception in ol es interpreting sights, sounds, smells and touch. %achine perception is the process by which the machine maps sensor measurements into internal representations of the en ironment. %achine perception is difficult because the sensors are noisy and the en ironment is partially obser able, unpredictable and often dynamic. %achine perception is the ability of computing machines to sense and interpret images, sounds, or other contents of their en ironments, or of the contents of stored media. 9eal#time perception of a machineLs en ironment is useful in industrial processes, such as assembly, inspection, diagnosis, ehicle guidance, etc. :ff#line perception of stored media is useful in medical and aerial photo interpretation, content#based inde&ing and retrie al of mo ies and images, etc. %achine perception includes, ; %achine ision ; %achine hearing ; %achine touch "ood internal representations for machine perception should ha e three properties, -) They should contain enough information for the machine to make the right decisions. /) They should be structured such that can be updated effecti ely. 0) They should be natural in the sense that internal ariables correspond to natural state ariables in the physical world.

Line *indin,:

Oine labeling is applicable when two#dimensional lines ha e been e&tracted from a two dimensional image. An important obser ation is that, -) 5 ery three# dimensional edge is associated with e&actly two faces# one on either side of it. /) A few rather ill# contained ob!ects called non#manifold ob!ects are e&ceptions#. The basic of the line labeling method is to attempt to classify each edge in the image as being one of, A conca e edge A con e& edge An occluding edge If a line corresponds to an occluding edge, It is marked in arrow. The orientation of the arrow gi es the side of the edge on which the isible face lies, using the con ention that the isible face lies on the right hand side of the edge, along the direction of the arrow. Oines corresponding to con e& edge are marked with a XH* symbol.

 Oines corresponding to conca e edge are marked with a XU* symbol. 5ach edge in the image has one of four possible label lings.

A!!um-tion!:

The ob!ect is in a general position# no two things are accidentally lined up in the image because of a special iewing position. The outer edge in the image# its silhouette belongs to faces which occlude the background. The ob!ects are of limited comple&ity, and particularly that only three edges e er meet at a erte&

5ou,h t an!*o mation *o ed,e +in"in, #+ine *indin,%:

The ?ough transformation is typically used to e&tract or cur es from an image. In this case it can be regarded as an edge linker since it groups edge pi&els together and describes by a higher order entity such as a line e.uation. The input to a ?ough transform is normally an image that has been edge detected# with a 9obert, Sobel or )anny edge detector, for instance. Oine finding in image, 5dge and line detection are important processes in many computer ision systems. This paper e&amines errors in each of these processes and how they are related, i.e. how do errors in edge detection propagate to cause errors in line detection. %uch use is made of simulated images which produce ground truth information, allows many trials to be performed, and allows analysis of ariance If we take a point (&>, y>) in the image, all lines which pass through that pi&el ha e the form y>Bm&>Hc for arying of m and c. This e.uation can also be written as cB #m&>Bc where we now consider &> and y> to be constant and m and c arying

(bject Identi*ication:
:ne of the most basic and essentional characteristic of li ing things is the ability to recognize and identify ob!ects. )onsider the process of computer pattern recognize a process whereby computer programs are used to arious forms of input stimuli such as isual or acoustic (speech) pattern. 9ecognition is the process of establishing a close match between some new stimuli and pre iously stored stimuli patterns this process is being performed continually throughout the li e of all li ing beings. There are two main approaches for ob!ect recognition using the shape of the ob!ect,

-) 3 i,htne!!- ba!ed eco,nition, "i en the subset of image pi&els that corresponds to a candidate ob!ect, define the features to be the raw pi&els brightness alues themsel es. Alternati e, one might first con ol e the image with arious linear filters and treat the pi&el alues in the resulting image as the features. This approach has been ery successfully at tasks such as handwritten digit recognition. /) Featu e!-ba!ed eco,nition: Instead of using raw pi&el brightness as features, we can detect and mark spatially localized features such as regions and edges. There are two moti ations for using edges one is data reduction W there are far fewer edges than image pi&els. There are three ways in which arious e&tracted features from an image can established arious ob!ect, <sing 'itmaps <sing Statistics <sing :utlines <sing 8aths

S-eech eco,nition:
Speech recognition con erts spoken words to machine#readable input .The term T oice recognitionT is sometimes incorrectly used to refer to speech recognition, when actually referring to speaker recognition, which attempts to identify the person speaking, as opposed to what is being said. )onfusingly, !ournalists and manufacturers of de ices that use speech recognition for control commonly use the term Moice 9ecognition when they mean Speech 9ecognition. Speech recognition applications include oice dialing, call routing demotic appliance control and content#based spoken audio search simple data entry preparation of structured documents speech#to#te&t processing. A typical speech recognition system consists of, -) The signal processing algorithm which e&tract spectral features from the acoustic signal. /) The language model. 0) The acoustic model

A-- oache! o* !-eech eco,nition,

-) ?idden %arko model (?%%) # based speech recognition /) $eural $etworking# based speech recognition 0) 7ynamic time warping (7T()#based speech recognition 3) =nowledge#based speech recognition

I!!ue! in !-eech eco,nition:

-) Speaker dependence ersus speaker independence. /) )ontinuous ersus Isolated word speech. 0) 9eal time ersus offline 8rocessing 3) Oarge ersus small ocabulary 4) 'road ersus narrow grammar

A--+ication o* !-eech eco,nition:

-) )ommand recognition /) 7ictation 0) Interacti e oice 9esponse 3) Automoti e speech recognition 4) %edical Transcription F) 8ronunciation teaching in computer#aided language learning applications G) Automatic Translation

Semantic!:
Semantics is the study of meaning. The word TsemanticsT itself denotes a range of ideas, from the popular to the highly technical. It is often used in ordinary language to denote a problem of understanding that comes down to word selection or connotation. This problem of understanding has been the sub!ect of many formal in.uiries, o er a long period of time. Oinguistics, it is the study of interpretation of signs or symbols as used by agents or communities within particular circumstances and conte&ts. (ithin this iew, sounds, facial e&pressions, body language, pro&emics has semantic content, and each has se eral branches of study. In written language, such things as paragraph structure and punctuation ha e semantic contentN in other forms of language, there is other semantic content. The study of semantics intersects with many other fields of in.uiry, including pro&emics, le&icology, synta&, pragmatics, etymology and others, although semantics is a well# defined field in its own right, often with synthetic properties. In philosophy of language, semantics and reference are related fields. 2urther related fields include philology, communication, and semiotics.

Int oduction to - o, ammin, +an,ua,e: Li!-

The name OIS8 originally stood for XOist processor*. Oike most traditional programming languages ,OIS8 is procedural ;OIS8 programs describe how to perform an algorithm.

;This contrasts with declarati e languages like 89:O:", whose programs are 75)OA9ATIM5 assertions about a problem domain. Oisp was originally created as a practical mathematical notation for computer programs, based on Alonzo )hurchLs lambda calculus. It .uickly became the fa ored programming language for artificial intelligence (AI) research. As one of the earliest programming languages, Oisp pioneered many ideas in computer science, including tree data structures, automatic storage management, dynamic typing, ob!ect#oriented programming, and the self#hosting compiler. 3ASIC )ATA (32ECTS- LISTS AN) AT($S: The basic data ob!ects in OIS8 are OISTS and AT:%S. OISTS are like ':(OS like bowls, lists can be empty. Oike bowels, lists can contain other lists. Oike bowls, lists can contain things that are not containers. AT:%S are like the things people store in bowls. An atom is a group of contiguous characters including numbers and special characters that stands for some ob!ect or alue. It is indi isible OIS8 has two basic types of atoms# numbers and symbols. Symbols can be a combination of alphabets and numerals. A list is a se.uence of atoms and+ or other lists enclosed within parenthesis. It is a con ention that all members of the list ha e something in common. LIS/ /RI$ITI4ES (R F&NCTI(NS: A OIS8 program is a collection of small routines which define simple functions. 2or writing comple& functions a set of basic functions called primiti es is a ailable in OIS8. The basic primiti e of OIS8 are classified as, -) Arithmetic primiti es /) 'oolean primiti es 0) Oist manipulation primiti es

/ o+o,
8rolog is a logic programming general purpose fifth generation language associated with artificial intelligence and computational linguistics. It has a purely logical subset, called pure 8rolog, as well as a number of e&tra logical features. 8rolog has its roots in formal logic, and unlike many other programming languages, 8rolog is declarati e, The program logic is e&pressed in terms of relations, and e&ecution is triggered by running .ueries o er these relations. 9elations and .ueries are constructed using 8rologLs single data type, the term. 9elations are defined by clauses 8rolog was one of the first logic programming languages, and remains among the most popular such languages today, with many free and commercial implementations a ailable. (hile initially aimed at natural language processing, the language has since then stretched far into other areas like theorem pro ing, e&pert systems, games,

automated answering systems, ontology and sophisticated control systems, and modern 8rolog en ironments support the creation of graphical user interfaces, as well as administrati e and networked applications. Subse.uent e&tensions of 8rolog by the original team introduced )onstraint Oogic 8rogramming abilities into the implementations. The general structure of 89:O:" programs can be summarized by the following four statements types, -) 2acts are declared describing ob!ect and their relationships. /) 9ules go erning the ob!ects and their relationships are then defined. 0) Vueries are then asked about the ob!ects and their relationships. 3) )ommands are a ailable at any time for system operation.