Ve aYVi’Vei a unjks3 mid ng i " 3 cept - ) Design finite Avtomata pith g) accel the Scrvings vohich start WHA 1 ang end: wish'o The set’ Of passilole Stings = S10, 100, (910 H--¥ » Met QE, SAF Y Onproain “ Z=EONY = %y ap dr | ay Le} do a) DL) Write: the RE ow Falowing. _ ato accepE the’ String Gook ov Book, ITD Qccepe -the Sing ecolos PCowuor CYT accept any +Vve integer Corn Oy) Optionat “decimal POINE addrers A) To checka.s-tying is an emailor not |; polling ETMOT Detection g Cosre ctor) = gemscireyr Sit Dfstanse mao wtih he Mininnen edit Distarce is ~fhe NO-OF iInsetio tetf0NS ANd Sobs-LHo-tHong Required -to change ’ one sting INTO ano-thor S-49ing Mqosth AA Tip -TwWO Stings XY OY ole = the Minion edie distance blo x py mZ— bength Cx) n Z2— vengthyy 5 for 1=0 +omdo 4 aist Lic). P -torj =© +0 ndlo dist oi Je; toy [=0-+0m do -for {=o ton do ist hi yJ= ming di's tliat, ST-Hinsevt ~cosk, \ dist Lisl, j-1JHubst - cost (x;y, ist Lid s]+aeetre “es EY aD example Foonvertt the Skving’+ctos' into Fomor’ Using Minimum, edit, distance -Atgonthino al i - Db TP MPN-edit- ‘djstoancee Vaile = Sl1s Meee VEC reghceporor] Peplace [error ‘ins evE InsevE | MINLP, yarrove insevte +l matyn Ef string -te starting nu (AF ex petali AEN ni + qa convert 4 I + nu}) nv to 0 " wom operators Ene converte the st4ing ‘pose’ into by using min Edit distance Aq ‘hovee! he min no @equived 40 Change qose +0 horse fs 24Wad RS ARS RAE 5 words are Classified into cateqorier catled Part - OF -speech _j-tpere ave Some timer catled word classe are “ etical cateqgower —> These lenticap Coreqorict axe USvall helt syntaHe ard Pp Morphological behavicoy —the Most letical common Careqoner cure Nous and Verbs: FD Other lence categories Tin ude Adfecty ey 9 Adverbs, proposiHone and Conjunc-tory > Word Hatses ave further UatsiHtied Into open # Closed Word class eg Hope word Clayces Constantly pohited closed word clarcer dont @ acqurre = Nevo Vyembers Nouns Verbs, Adjective Adverbs Ane Oj Word chassey | re —) prepositions, conjunction , -huviliaxy verbs 2 Particale ane claced Veatbword Claeceg > parts- of — Speech Example NN Noon Conair, aac J VR —pverb Cru Sow S40 4 pa 7 (lowge bfqin, Patt fers) AD > adjective Uae Wy, care Foul, uniformly’) 4y —>-ddvevb (stowld, to, oF) EN —> preposition Cin OM AE 11 pee > pronoun Cre! Ne pet —> Dekerrriiney (oa The INES those) part -oF- Speech Taq ing {6 the proces of Assigning a part -of 6 Peech (Such a4qa & Noon, Verb, pronoun (P* rue based taggers used handcoaded autau-+to aasign -Laqs 0 word 4 > thee wile USe a lexicon-to obtain a list-of Cardidate Logs & a use wilet Fo discard incorrect tagy Rute bared +aqgerS hove a +00 Stage Architectey he er a 1S a dicHonary look UP > cedure, whi RetorS A oot'y, P “bags for each ieee po-tentap —3 The second Stage User aXe OF +ttandcoded Wley Ho discavd weorrect tags +0 qeea sing) & pant of speech Fer each word en” considey 4he Novn-verb Ambiguity in +he following sentence “athe show Musk qo en” roe Retolve this ambiquity by using ashe Following Rule: ATR preceeding word is determiner THEN, eminate VR SIF Word end in ing and preceeding word fsa Verb hen label it ay a ve. Advantag 4 AANA no speed iS acdvartag eS ee EB More —Hre is required wooting a wile cet Dit fe useable tos only one re srocharHe Toa" “umeadoeaes | moder] andard stochartic. +294 ey 15 pene (0+) ane Most liter sentence we —pThe st the HMM Tagger’ arsiqns te9 sequence +o a giverCaty pig He 4HVIM because, Fe OSes aWwo ‘a4, : Of state, . } J) Visible layer which cormponds to He ilp br QV) Hidden laryes which cocserpond, -to he ilp tag, "Giver a Sequence OF word: the objective sf) Pind +he Most +o4q sequence tow the i Sentence: Lek w be the sequence o+ words b= M1\— ty Aine lark. 15 to Hind the -toq sequence 4a 4, -- tn Which Marimiz ¢ probebbili te of the be =~, \ T = asqrnay TIlw Applying ihe i 7 PCT ? Navie Bayer mle p(T lr)can b¢ BHmated vsing -the eupresion Perm =P[22 ] 4 ot] H Pl] Ne €EXPTUNON toy tre Mos © Wheely + Sequence become —p! = axqman +) 4 1 + PCH) PC 4 being +he Mageov modes Reercenee +he D ; probability of {aq sored gs CaN be ertimatel at the psrduct of prbability of i441 gram, | PET) = peer) Fp (ta/ty) ap bts )ea) -— — a Pln [tt ---4n-1) plvir) fs the pwobabili by op word Sequence | =. €%2' “athe €qq fc Rotten” Given “the” speci-tt e¢¢ the. per, ‘eqq"5 Noung fs —> Basic verb -Form : [DT NNP ve tT). Wem axswumptorrs|. The wordt are independent -to cach ons asthe probability of word 15 dependent only on ity 40g . using —thele pesomp-Hons We obtat’n plwit= e (wih) * p(wo)ea) - 2 —-p (wi L4i)y__ p(wn }4n) i-e---) Pliotr)= ot Plu! 4) “peo ereD~ f iP sib Daep ber) Poh pl Leg Jepte)-— 9 Phen bing p(wir) can ve erlimated a4 plo} PET) = f pwr Ha)# P41) Fp C2 fi), Fe - 1 p (4 jure tt) ies 7 pe etna ted this proboualli al -foom eR lative fesequency vila Maximo licety hood eptmatior pAthte2 ete) ~c Cr-2,-+Ht-1 4 ) Eee e(4i-2 ,+f-1) P(wikti) = eC wi) c (ti HY) trybad Tagger Method ae Hybd reaagers: Combinen Ahe featore of! both the awle! baie ip gtochar He “Approach The Transtomator bared tearing (TRL) b+, Tags fs aN example of Hy bidbnanno +etcd |. data stem Zn ernf can? daba & A ‘n-Hat Stade —rimo+ation — Step-12 inpuk Tagged cospus 2 [eticoN Hrabailed evesy word with a most lileery aq f-peorn dictionary} Step-2 tA check every possi te ee eg @ Seteck one pohich im PT? a toaatn, Step-3 2 Rerog Cos vs apelying the Rule. D¥ Repeat Step -£% oni l stepping aderian fs @eached Ovipot 2Ranted —geqvenc © OF qronsfoma : UV, hay) Saptege ARES [ene per »] 4 cFG a: ConcHty ancy s parsing CR&- Comient-free Gwarreny : aaconsul he Pars 1 glep- 2 Ex: Reena Road a Retna SPl es» up vp | aN Vv —eeenav NP —> Reena Readi Dek ul J peena geadt a Book che, Pl by ana cea cetination S—3 Np vp | f* Np ON. , Qe Fan Psy N rohere P Dee N = start of the Vp OV NBo ager i set | Vp OV witindie | Non Tem ina] consttoenty 34 Ce "les - No-of Tevminaf arse ‘pe ' production wuleyGemuping of ditt word grouping OF Sma, Word frtoa lorqer words: 4 this 6 Above tHe congo €x:-the bind fi con be converted Vv Yrto 4He cosveck Sen}, cotbergpig Pe ve toNP) e brrd can -Fty |eonsHtoen em ¢ uf [eonetttecrey) So wile. lagen Unit oF 120 1dy L) phate s4mcture wl 2D) Sentence ctwctore wily Conctttu en cy phat Vy cons tecHm contence lVy v i : construc Ho Noun phyaye wily aaa mw Np —> pronouns combine au! nf PP Perased henhwt p> Pet Now converted into Np 3 neen Np alper Cady Np —>-Adjecti ve Noun Noun |prenown Np —ppee Ad Noon J y Wwe Add 4 prepasitan hen we gee pip —> (pee) (4?) Noon (pp) Ceesaaa rire) vp? ne et rnin vp-7v NP vp Non” vp— v PP vp Vv NPNP emmy aigena Vp Vv Np NP (rex ~pundavacho ondata Povachu: Erircthe boy Kicked -tre bal) _Anitha slept in +he garde Anitha slepe prepositon Pores ® Le Stawts with PP towed by NP he, pp —> prep (NP) eas you played votteoal 0° Sine beach il) pp 3 prep Cond eee pp) ens He went outside Dajective phrase AMY ry The pp Consists of Pdjective vohich be pesceved by -ine Adve to and followed rerded by Pr posite ex: 9p > (nd/)-Aai CPP? My sister iS found & RAM alk he ain cig. Mey \ate Ady 0-5 vex PRISE pbyerb prraye —OCSEn tens )- av (Hdvp)© OE XANP th x Prreceeced by sHhe deqree v . (aviqh Higher Hihe Extime passey veny Quick y (deepresent the leg, Sentence Sf-Compan Son MAND & levet Cons-hve.tieh bs De clawative &: Imperative S Yeor-No Qvet-Hor cyrvc-tore Us wh- QuetHon cyawetore ve Y¢— No gveHon o4yy0¢ Fore. ve Ns Iw SHOX Ne vp EX: Do YoU Have common Sense LS Are clats over be you have black pen Can me Show your Notebook wh — Qyertion Stroctore HE hich courtay won she. Highet gotd au-7 Vectalt ap bohich Camerar can: you shno me \n Your Stodio (thop vy ; SFWh—NP AUX NP vp COMAINATHION Sentence gad Vy US Tee Apple Garang (Noun) bettoeen ether Noun over PSpie is dazxing and palnin : bo petrwoe en Veto hi® 16 QHEO ANAMOCT CosHteN eo LOVE SiCto% aynqreem ent : oe Pocs priya Sing vy Novn( 374 eee singulan) tren we add eitnen 6 oF €5{ ver Fi) Rqgre an ne wie S340 NP vp SDSSqRUX SS4NP SANoN 264 AO Non 34 NP vp ENT SSqhex D poe | haslcan Neon 3 sa-ko% 3 do| have} can te +u WAM le Roe LE is nothiny but get oF #eatox Vatiue Pol ee I ie CAT ONY? j } >) feature Heme wg PE 9 crructere peor a ag reengent etsuctu ve CAT Np -Agqvek Number re row pexorm 3[ony favre Pe) = f pwn PL] | [Nom Pcy_j[vern c9) > sen pt ] [tem pe )e pea ser] = error Paring DS theve ove Y -Lyper OF paring Fechniaey t)-top - Down &) Botton —up 3) eorety pawing 4) uN panting . e Easamfh Top-pown OF Nearer ww NY St4ra4eq 9 fo a parsing Stxucctpre thet begins vof-in the start symeo) and attempts +o dewive the ilp sHAinaby Ppp hy G2 the Produc-Hor) sv ley "% Steps in Top-Down Parsing i Ihe paning procas begins with tne chant Symbol of the qwarnmary F the parses expands Non-rTeriminal Synnbo Ls by Ppelying the Coneponding producti) rule 7d Top down panin S the parse +ey to match the lp Stang 4, With Ane Jerim Syrnlool Ve Lee poser con Merch -the 1 /p S-hing te ry Ch rack and hy diggenent Preduction wo tes 'paged PACS? a pet Us e he gentence “the cat -chated the 1) Setar of Hhe symbol a 9) SNP YP ve 2 Vv NP. Np pet N Vv crased. pee Othe NPS peeN Np cat pet >the “Top - Ogyon parse “Tree N tae a) Np vp Det V vo NP sy i , tne CAE chased i pee N Bi B the, FOr Reto1g Op parsing “Tess J itts.a SHtoe9 4 Symbolsound Symbol SESS ORT, BA that-Start votth +he FP | attempt Hp dedve He taney ine pares 091 begins /by examining Ane vfip s4¥fng om) ete 0 Reqnt | satel ok 2) VC identieie re subctaing ata match the mond aide of a produc tion PUle ed gerbsti ny ane replaced by onding Non—Terminal Symbo| ave, ure tne fp omhgPores Ing proces ] Whe cat shared Ihe vat 2) she Dat N verb deb Al \ne Yat v avy Pedivod ~ne coho lIHisa dynamic programming Pigovithm Used for PAWSING Cen-tencer 1 cortent ree qvoroma . Tt is paelicolaaty Use-to\ becatte bt cin Handle au -typer of Corrtert ree qaamman > Carety panin fe named after Jay earely WhO Yriroduc ed “i IN tqro's ‘ Characterstics oF Carvery parsing bette handles au conteut fee qram mam \ & 1t--follows “Top-down —Apprvach predicting ; Powiole parsers ee ; 3S e-vuse, . dynamic programming —1o ‘aNnotd Redurtont Compu-ta-Hons u. Carey pastepets USeh -Hhsee Operacttony : [: pwedic-Hor) : Q scanning i <=. comple Hon fp irin, Artgori thm, eae $jst a iA ttate fo Pepresen-ted of A de. wee Bytn > BL IND<-Bi? Lone re ans A—>«p, 15 a Production Rule inthe 7 indicate yd ammnay 7 Cureent “Pos HON IN =the prodve-Hon, e anh 7 Rule TD position tn rhe i}o Whore this oule sharp. 2: Chast + the -Prigoristhm Main-rains o éhase rhe, each entry COvre\pordi to a posi-Hon IN Aheilp tang» each entry certain, A list of statey CHeps, IN -Eowely parsing ‘ Be I. €arety Parsing Wvowver procent the ilp string by iterating ove each postion inthe Sting and -PPPly 3 Operca-Ho ny HHS Step adds poscipte euler +o the ehavt besed On —ne Comment \efate -(¢a cote @ empecH a Hon—Téviminal nene the paweor predict statt Non- Teriminag production Poles Cp xediction ) Scanning : This Sep Mactche ne Next ilp Symbol with -+Hhe -Eupected —ésiminal n-he corrent g-fecte: |-f he —teviminal matche the state is Advanced by Mioving Fhe J Port the Teximinap Completion 3 voned Ci) Brees Eres of —the prediction —the £ ; Bi. guts: -Exampie:- Consider ane ile: pare eS pioscle pate pnt rhocota-e 418 Np vp NPN ’ Nip > ip pp N—ppPaPpy op > p NP y-smillk | VP —>VNP’ ni —Schocolate | — drink Vp>yp NP. VY 7? sith | r seposi ton —> with (so) Tt Ga. [oie — “tayd) fad (uy (U3? (> BY (S105 | [sx ”? pp>-P NP] pai powrtth to) Ten Be eee NPS Np pp |N-Smilk NpON* ly NpoN. ; Nill fup-3 we-ppI Co) au Cet as [PV NPL opin vooarink vp >V-NP pac. | : 1 (0,0) ‘0 Sian Tee's [AeSpapy (OO? fea Nese eb] S2Ne wP = INPOON ce NIP Np PP] He | otaa Noppapo he 4 ABIL SL gy i *_ onthyo / athe oye Pigo sh fe a Meyvnbership’ Aigosithm Used yo creck wheather ihe tp belongs to the. FAtnmay ov Not x mma shovld be exp wes tn +he orm of cur gare CN Covet shovtd be ‘ DIBS : AoA : craracdersttict ‘ jhe grammars Mut be C he-fo ve Applying CY -etepsinn athe pigosi-thm constree + sthe parse able in « a Bottom up Manne ii ky ; ; Z.jt vse a Table +o .gtore \rtermediate Redlty pohieh helps in “etticienty parsing the. i|p staing Steps in Seat within my |. Convert Garman 0 CNE 2 Intatize ne parse tabie BS fit the sable Wo Hechect tow “the cto syrbo | ies \ onverted —toe chomskfin Nowmal fovmFRapabtlette pagicg Ste Ts an extension of coment Ree Grammar sy) The dyporet probabilitet into tHe produc-ton | te is used -bo handle ambiquity Ard once tas) (Pere) sa Prohabilistte content fee rere fe a cre in rohich eyesy whe js assla} eo 4 Probability. He extends ine CFO by ary mH Lach wle- 4 ADK VoD ndi-tona) IN a Set OF productfons 'P', vabsfeh & 1 Probability pw xtp) johese 7 ii din. a. constite P= probs ity of expan q iu L579 Ae wle +A—> % ant | | A PCFG Po defined by +e palit (Gry) Where "eis a CFG ‘ ‘E* ¢6 the encfan y | ¥ pefined over the set of wulet Such 4hat-the | Sum of -the psobatilt iu associated With the | wuoley exp oot anding a partcvias non-~Terminal is ZFC 90) SL 1 SD Mp vp Co#) — pp->pveposttion NP CIO) | S93 vp (O2) pet D445 (02) Np pet Nown (oY) Deb that [o2) NPS Now tor pee 3a Coras? Np > Det noun pp lo-2) pet -s-the (0-35) _ Np Speonoun (0-29 Noon >paint (oa5) Vp > verb wp los) Nown > doot Cas) Vp > Verb (03> Noon 5 bird (0-85) vp—> VP re (0-2) Noun -> horse Co-as)= po leepslo'2) preposHon —2 60 1(0-2.) La yerb 3aings (0:2) preposHion —>-+0 (o-a 5) yore? OPEN (or2) pronoun —2 She (0:35) yew? 3 56aW (0:2) Pmnovn > he (0-35) yer > paint (0-2) Pxono0n — -Hhey (0-25) eposHion 2-Prom (0:36) preposition F WHh (oan) 9 Now we can venty hack or each non-Tesiminas Ahe sum of pevkability jo’ 1! 6-3Ne veH4-4(65)- O*+09 =| ' _p (pee -> AID) +-# (Dee Strat + tlpet a0 4 FC per the) = 0° 246-2. +02540.86 7) estimating Role pwhabilitiey rr ; ey yang Maxime li ketihoad estima Hon—-fonctir 40 aprimate A prdduction Role AO foqven -toy the expression psobabili ty ot max likeliboas Pave (83%) = count et) \ _——__—_ 4 count hh) x —> the pecbobility OF aComPUTE pase Ee is catcplated by Mortipying re! pobolli ty or each of ihe WE Used in Fnewrting —the parce—vee is defined by +he- Probowsli of sh POS) = pi) re forse aree n= node of the =n oo wile +0 expand.n g =sen tence IL w probabil kebenseeceoch Pate sage_— —— lUpy p - hole EX: paint —the door WIP 7 s Crean vp oz peu * Very Lip es | vp — on LN AN pr On PL) = (oa lb2 =) los Cou) (0-85) Coa’) (ro) (o.as: we b>: / \. a deo “4 re WN | “035? (as) wT p Gx) = 000000 la) pee ae : | Deb JN ane of —he') hole 4 for a qven Sentence, +00 Parse Tees ant | Possib lewv, se " sobabili FY of “wo parse -eeet Cf +he eP? lence “paint -tne door WHh —the hole” 16 computed & —folows “ain —Ihis example . the 194 parse tree hay Pi pnese povaisiity teasing +o correct Ie + ere Eatt OD we can catcevl ate —the proioabil wy Summing UP Erxobabilite, «oe awodlated ber 25 Now: of Bemlenc ey of all possi ple parse Se