wap Cross Demoin _Knobdye Coe Concepts of Poject |) Named Entity Recogrition fuer) proces of Binding ond labeling impor tort pee is the re ive“romes oF people, Places, Hiegs (entitier) text, does, organization’, <* Be: “Tet “Elon Muse Bund Spacek in 200" Hee output = Portes -Elon Music + Organon tion= Space X + Date - a002 3. Relation Extraction | once wre Pind entities, they are connected, lation extraction Lindt has Elen Musk Bunded SpaceX in S002, Entities, Elon Muse, SpaceX Relation: Founded 3. Knoustedge Garaph Basies A vrowkdge graphit > network usheve catibes Ore nodes, cind* relationships ove bhy them ore edger, Tt helps Sere ord bok Lrowlkdge in 2 SPUtured wey. Ee Elen Musk fandd SpaceX plode; Elm Splusk puede: SpaceX Edge? Amucded @ Scanned with OKEN ScannerSueple Cede AILP Nip is the red, Understed ard emote eld of Al that teaches computer ty import Speech framan language jt anctital ze LC becoge 5 Dachine Wansletion ; r 3 Aratomarically trons lating tire Rem ond brgvage t [site cr Mice another ing At pine" audio. 6. Grammar Checkers by 4 ete 8. tecog Thole thet ind ond grammar, Spelling, Gn} panel You Punctuation €érors, in Ext Or 4 except: Example price (> Sovey ip: “She 30 school every doy,” Conmcted: "She geet t cchoo! eveyy doy.” wuloeles. dictation: Converting Spoken lagg” 19 tert Sitople Loder Nips cbse of AT Process of Dictation ae Capers (eacords your cpeech) Featuce Evtartion (cantert® oudip wove inp aumeral Y eo ba ve Aryattth: medel [Acoustic Model) TE Vvecy og Mops the Lengua mnadel(germaice me Poites BE Sond rm a oe tikely inte Phone(s > Pregl Thee Oven ( werd cpuince) SPEC Condgh a *C recagnized isord © “are Combined ard oS ritten tent Aisplay ed @ Scanned with OKEN Scannerches computer ty » language, 1 one brgeage te | Stelling, an ) ey. al eveay day. \ owl < Speeck) ourrerial ) nee Narecs) & Map § 1h lap The Son, Pots Bain fo us inte Phenomet((penci Speech ealiasic combined and din wove op © are Lo vitten tent. import perch eration jt initialize vecg)nizer pe CeBeQgaizer() Lath co. Picropherel) 0 pent” cpeate Angling: audios ealisten (Source) ty fete ss eecagpize qpegle (auto) prae("Yeu cad + $3 Beerat(tere)) cept aaa Meee Sony could noe vecegnize whet you Cod’) Vey issues to NUP 1. Arhiguicky b Lexical Ambiguity Single word has multiple meanings Ge Bing np Syntoetic Ambiguity Came Sentence com phtace tn multiple mMerings > Cymantee Ambigui Stee a ed hog) row than One mean ten words thenceloee hecaute of the meanings of the Prenmantic. Ambigui ee OTE na enconing 1 nc lear cox hen ‘ rr anton’ Mlention; OF Staton ey Specter @ Scanned with OKEN ScannerContest Under sto: the obilie words, ecko, HP te ie Pret the moaning wf Tentemess, OF COmMeEationc bated on convo, Continent ond eenotion conalysis Sertiment: Procas of determining hether 9 Piece oF beat eeprrSeS a positive, O2zebive or eutea) pinion Tdlentidy ond ch ssifyicg specific P metion AralySts: f E mot I emations. (42: jay, anger, Fons) Modules of nthe Pegject 1 8. Medel Training Date Collection ond Peprcetting 3. Lteb app ond APL 4. Dashboard and Admin ig phorekgical Level preourcation ond Sound at dele usth Cheuld differentiate bly Pet od cat Lewt 2, Morphological level tudes the internal Structure of the goed £ Es" Boppynecs bevel 4 d level ® bevel 4 ; bowel “Token 20 The pr Loo rds, Eo @ Scanned with OKEN Scanner9 Piece Neutra) opecifig Feary... ) | level 8 level 3: Lenical level fecuses on Laerd teaning Exe He went ste the bane giver bon! tactic level no hyzes Semantic level!” Focus on The meaning wf the Gentance level 6 Ticchosure level eve level 4 Ferm 9 mntoring Bll conver snfen Pragmontic level” : Understanding teal-world Knowledye text. into Smoller Unit! like Tha process of -cplittins en tences, Log dc, phrase. OF Si @ Scanned with OKEN Scanner4 processing tee finigue in Natural rg (Nev tahere word ) by choprin: oft’ root fom (Ste prefizes or cuPlxes method it often Produces Hote: sre is a Crude aI Give Cook efor) thot orent * x: Shdies > Stud? Lemmatizatt ea mulher tere PDtepeocess technique in Hep pokes sede ove vedeced te trae Gate or dictionary, em (lemme), Unhive DiPBrence bhe Mb ends PL Mi veA Cobcet of Al vahere algocthm'e “Le ‘ Kom data to eat frateeng + Lecce Vell csith Smaller data tere Needs ~feature Eqgineeting Bore on CPU Relatively Bast if Eater LGndcistend @ Scanned with OKEN ScannerLAt Subcoe of ML thats uses artificial eure! rebsorks sity mony layers to fear Complex patterns 7 . + Reguices Lye amounts of deta, te pectorm usell No need of Rate engineering, leare cutomatial fics ens do . 4 rod. hots + OP ten vequices GPU / TRU: | + Treining can take hous! days « Hard ty intaprer ein Dale or Me Cupeevised Leoering «(1 knows) Jrommar Un copervised — Learning ( eno) iS ReyBecernar beaming The model learns by Ww tetacting ith on envionment Fe Ceiving seuiaeds or Penalties , foe yes actions, and Wmpeving Though tial ord ercor. “Ee SA wbor leaticg chee — it .Z0to-g reread Br Correct move and 2 penalty foe & wang | move. @ Scanned with OKEN Scannercpeech IID) cech word in a Sentence a aby ee pore of Speech 9 tpranclation 3, Machine 2, Charbors and virtual acsisterce | ap He Pormation erect | 5. Sentiment nabs steos Noun- nn (Singular) , nns (pluro!) \ Veb- vb ybb (Verh pose tense) \ ot Adjectwe- jy Stedlvech — ch --8 Poreposiion = fn iorun'— "TB | Ntoieed Entity Recognition | L prees of jdenibging & clascibjing real world | entity “Ucdl Be ettacting ib revaton a > Uced a Obathow th reeqgnize Grete Sek cettivel in in business onahyties > Uted @ Scanned with OKEN Scanner9 Sentence Pech cal world ame of the People - Git Geo Political Entity (torations) = ORG Orzarimtion Be dtes £ } Dare —Money + Tt | qpLntedaction the Groms LAr Me Gram 1s Sequence of No Cons words mo tent, 2 = Onigram (ey: coords de f= Rigrom (Has)! THE contecutive & ee Teygram (es): Three concecstive Lorde Eat .e xn oe Sentences "1 tove W Unigrons: [12 teres Bagroms [ (+ ove), (Ca qd Tragrome:[(0, ove, N), (tore, ME), (Mite ay Lihy N-Grramd are Tportent : 2 Text Prediction + Machine “Trenslaton Cpeech Recoynitin J Seren: Analysis Te Retrieval (Search ex nes) @ Scanned with OKEN ScannerMotkematicc Rehind N- Grams . Peobability .IE- USieg N- Gram 6 Bigram NTo de! -?P [rs Ww : Probability of Previous Lord. Sentences acy, peinjey TP (vai fu; 4) a word depends only on the @ Scanned with OKEN Scanner