SHVESH

NLP ASSIGNMENT - 1 o3S|||621

AIL ML (e sEm)

Al-4) ouputa tiou:& ghonologa is a fatcivaty inteoseetion of comp-ts

sciemce cd lhsuishes

olonc cf copleK ds,

krdepnd aut Stas: ie morpnem es ton ctanal
omy hci Oun mmeauig. ihutge

odrrbs.
deg as houns, veibs, cdjcchvs, r
Bond ohenes:
ocrtate a
weed to be atio oncd o otnes mophemee (usvlly fre opncnes)

(Ò wond Eros! These inuolve Mistakes vtu athvce wosds 14 a seujnrt. Mey
hsgpen wheh he choose e wouç od or mss pell a wod hot exiss jn the
Xamnplet:
"SubskhÄOni Repleciug one wod ita anoer.
 mtta kes

ar cifeufially
ERampler
Typo vi no ral- d tquvaltats

eRaieunt Oud warkes ell foY larse datasetr. Ihe trsrde Eag

morhotasoal calts Gud uses adtomer to ideatiby the base or ditoverg

dSourst. t ondlypes how sataies ar inhed tozctrer to cocet atoulng en

meaniug bl hole. Ekamincs Lonese derives ine phonomscon nothons (aud, but
belaute) nd icierae mechsuisms (ancphoa, ellipsts) mat eusure smobh traunsHous
btheeu 1deas.

) Namecl Enfty heosutoy (NER) plans a thcal nole 1h MLP b1 octhg as fe

dott. Hor's btcakdoum of s siquihoncei

NER aeS ino he vast oceau of fxt od fi'shes Dut cpcel fe namMed
euti Hes, tRe feople jorgoul 2ahons, focohious , dots, wgnetarg valves, and oer
 lox NLP applicchiumt do undeotttmel he coutat bnt meouiue of he Hxt.

" ohancing NLe Atewsaly

exeme : Tmasine a kem us te trseuce of a wd itt <or meaug

Ane 1eKeme. This en0mpa sses the base veo "wn" jself, bt alo s nfeenons ee ns,

in almge- It Can be

Morphemes: mee can stand aloue 6 ume hioned words or coukut words,
e must be atta ched
Oud o

MoYphol ofg : It (on be vitued as a mechonSm for 4ppmding

() Iutechoned witin a
mes to indicate thclr e
9YAbmmatel menns to ex\shug lexe morphemesy do ot (Kat hew
Rhgun as fntechonel
Sentnee, These markes,
Yaner mocity the 4Ya mm atcal fneton of the base wYd,
wnds but
morphemes enode vaious grim atial Cottso fes;
fnetondity: Tafte cional ksl, ualkcd/),
f
pregeut, e (eis aalk/’ pal
Tence: Past,
Singular or pluae (ey lot/’/un/)
Casei Noni ho ive , ateuSative , eh.
Peton; First, SeUOud, tid (e I nn, you ronl, /he ms/)

offen epejautd Usig ymbo

Techvict Repesentution: Fnflectonal morpheme axComgle, ina le-based systen,
oy hotatons atathed to te base leke me. For
tne additou of tme plel mgpheme s" to me vts
 is an eseutial task fn Naal
(p99) ta0ógvawmmaial
tass (eig noun, verb, adjecte )
La
frocsng (NLPS tht astgns
each worid In a seutenre, Hex, e exploe tree Rey ppoacs to Pos tsg
wle-based, tanskmahon - basecl, dud staisial.
Rue-Based pos Tasslg- Tmegne o metiulas ingoist mehulovi'y ngig
Sentnce, Rule-based Pos tngsis adopts a similar appoath, rying on a prede fned set
of hond-oafttd wis
assisn ta_s.
as a mon flexi ble

by le-based tans fomaos fog chhementf.

SSahshot fos Ta_glas : statshel POs tassry take a dota-dalren p c
fom lasge amouuts of pK- tssged
text opora,
H2- NLP is a ffeld of AT at deals bit te
ntta eton befueen "p
ohd human lousuag. It eguips omputn cim me
aud sencte human (oyguage in a abfihy te undetoud, iatipt
NePs omppntuts aud te ellteytut mtanlughut wag ttor's a brta kdo un of
pneses faold:
Componeats of NLP:
* Nava Languase Understndng CNev): üis compoycut fowses Ou
enasiug hachines o tonprhend the meaning of lousuage . It avolves Tsk

fatof speech (P08) TAseus: Assi_ning svommcheal taps s each wdin astutatr.

" Natve longua hen eratDn NLh)? This co myonent flps tme SHpt, a'loeing
Gomptsto phodue human -KiRe Têxt. Tt iucwder tos ke ke!
oMachine fanslatoni Comrertins tent fyom one tousug to aupmeo.
TOst Sumumas 2esou toudlnsla (eug ay tat indo a shomraAmte smtnelg

ConverSuon it humauJ.
 Phases of NLP!
I. LeGe Analysls amd Morpholiglal Analysle: his tulhd plose brca ks dpun
tRt Fnto ts basic building tbiocks,
"lexlue Amalysis sesmeuts me text futo wrds 0T htentes.
of wds, idandkyhe
* Mpholyiul aualysis examins the tutmel sthe
Morpnemes (ne smaliest manms - camglns vnit) ad tmelr vatataus (eis,
"plog "plaqs", "planuy" l skm to mthe Same morphcme "play).
2.Syn tache fnalysls CParsinç) This phase Focuscs on pndenttndng mec
foamadttied stoee ofa sentue. It analy2es hou wors ae
am tlated to each 0tmer

ancepts.
4. DistouYSe Inttsateni Tis phase fowscs on tow sarttntes ar counettd
to kom a wnerut disoute, It analy2es ohesi deviies tike pongns and
onjun chions mat ene asmoo ow of deas.
S, Prasmatie halysis: mis fnel phase dels nto the deeper mconing ano
be hind the lansuage. It constJes fackors ike fne touttt, splaheg's
tme Uhdetgins nesoge.
backqund, and Sharcd Rnowlcdse to understnd

A3- Tsses and Chollenges u developln NLP System i

Ambi_ulth: Navee aguage is tnhcoutly anbigous, making it cha llegly
t. conttat áecwrately
ko machines to unclerstnd
acos slons, As, anal fudiuidut
2. Vatasila: Langag use Vaes dely model.
naklns dSeiwlt to Crt a ehiverul

3, Dota auallty Np Sysems guie vast amoun ts of higa -guallty data

acwatly
for aliMs, hich can be ditfwr to obtajn ed wotute
acooJs feat domain
t. Domaly speißthyi langvage vares gaeaty
tegel), equiing speiolil ed models for eRectve poieJSing,
tej medicae,
 he
TAing doa, teodiug to wnfwy oy necwnt otOmes,
deislos is wde for tvst ad aco wnta bility but rmalas a
7. Molslnsh' Devrtopts hondle multipie toasvago
NLP sytws that (ou
ekechely
8.
adde compiea dve to ls vishe difeymcs,
Poivalg dud Sewiy NLP

Hee aye some

1. Liee Ambiguig
Champles of aabi gi kes u natnt (ansag-

wO4$ bim multiple meauings, rample bauk" can mean a huovclt

2. Sgntockic Aonblgult
Difercntf fnterpta Hous bascd ou the stvefvn of oa
I Sau the mon Lith the sentne.
telcSCope ". (D{d I bse he K|escope 0r did
fne man hure it 2)

3. Recernt al Amblsuity?
Uncetint abot what a pronoun 0 noun en to. Fxample: "She
told him that she loved hlm'" (hiho lDves chon?)

. Senante Ambiguity '

Words or phrases itn unclear Meonlugs. Exomple:
he chickem J tody
to eat! (Ts he chi ckeu ttady fo be eatea, r is it aody ooked 3).
S. Prnsmatie Anbi guity
Faterprctatous Mfncd by Lontat oy spla hey lutent. Erample Cotcl
y9 pass e saltl. (BIt a dtguest or a qesNou Rot ailig).
7icy use technigues ihe:

Phsed dRetty ran m lukomahon you seek.

methins to onsider Sjytn
Semautc Seanu honng beymd eaact keyLord

couttt of the
hal te
Lousideas fackoy line he con tat oyo ueg
cbpoges. yoch pne

Translok. These tools use tcanifus ihe:

rusloon tools Iske hoesle vast
Sfatis hiot MochiMe franslateu! Anal2ing pass
'aCOSS
tlanon ships beteen ds nd
Aota fo (ean the statisal
lamguay. Uhliiug cleep leany nodels tained oy
Nenl Machlne Ttnstatoni acwate ad huoM(Cel
tnslato4s.
dasset3 ac4ree hor
mossir asto mct
bom Sixs nd Meka to
Assistdtt
1. (hatbots and VýNl

extcthag key mkormaoy ke home),

Reoghiny knd
Entity xa honi
hmpt.
teatons or lats tfom uer to uses ua
hael, hubanr llhe
Nanl lengase heneyatoni Responding
cpnrhathon.
tajlord to the tontAt o the
vay
trxt ) chctaQ
Unden nding the emososl tone of
putaon manage meut,
for tasks ike socde media honi toins bibnd
wsfomer ked back Qnalysis. NLr helpr byi
A6 l. ELI2A
Eliza is oye 0e bhe eaist ekompies of otme tanguage procsia Omal

Cnd simple tans oTmotHoy nulS, Dt acted as a psy chothcoapls t,

tugagity
Sesslon.

questions, EL2A rcatcd he lluslou of undevstndiu ad epay Ospite

it SiplistHe oproach, Pu2A as sp Nsingg efective at nsosinç uits
and spartng meaning fl neyactlons.
2. LUHIX

LUNR( linsuls8e Unive seb amd Noved hoyshe Relafous) is apeject

that a'med to deviep a computr yetehm epa ble of underttdig and
gencratng matrl tamguage. Fnl5a4e4 ta tne tat (970, LUNHe foused o

on (eamed patems.
LUNge contibuted to tme inders tamdilag ot syntau

bsed to bld compotatonel models or vnderstmaing and

humay - level language undostad'

NLP.
summayes hat copre he haln polutr. This is uscol oN applchnr
(he

vNeul aS55ators: ouldinç tonsize sippet ef neus asmeles.

Dowmef summazafion oott: atng sun motes of rsloh pape5,
lest dotwmeaty or emalls,
a ouvet athOn.
° Chatbot Summazing key take auuys from