* NLP *

Chapter No-O4 *
*Semanhic Analysiy *
*Amwen he followinq uestions
C4 hat is SemanhcAnausin lhy- Semanie Anlusio
clifficult 2

A *Semanhic Anuysi
C Semanhc Aaulusin Hhe proccm of hindinq he
meaoia Erom text
of Semanhc cunalusis o o draw exat
C2) The Cuim
meaninq Or dichionany mecaninqErom the text
C3 he most_Lmporkant tczak of Semunic Cunalysi u
g e t he propen mecningof heSentence
u e c in extruckiaqimportant
C SemanieAnadysi d
romachieuinq humao level accuracy rom he Computers
o _ uned in wola like machine Eramlaiona,
chutbuts,
C5)T
Seanch enqineD_nd Fext CLnaysisA

Analuse heSeatence Chouiad i qrecat

Example
Tn h i Sentence, he Speakeni Ealkinq abaut Lord
bou whose n a m e io Caoviad.
Cnovind or C penson

NLP up Connidened a clithcult problem io Compuken

SuencLe.
he nutune of he human Lanquaqe hat makea
.Tr's
NLP difhcul
The rules ha dikake he pcuoinq oF inform cution uing

natural anquaqe ane 0ut

ecoy DrComputens u
understand.
2Explainuith Suutable. examplea ollauing Yelutionshin
between word meaninqa
llCD Homonumy c2) Polysemu 3)Syndnymy
CA Aatonumu C5) Hup.ennom 6) Hupo.numj
J
C Mero.nomu
Ryponymy
AoA
IE io delined c the relakionahip between a cenentc term
Cand iankance of haut cenenie term
Ferm i Ccllecd hyperny.m Cund i t s inatunce
C2) he qeneric
re Called Hypony.ma.
Ihe word Colou hypenoumCund h e olou
Example
ed 9reen etc Cane hyponyma

C2 Homonu.nu
C) Tt i celined a A theworcla hwiaq same pelling or
unreluted
Sume torm but huving ciffenent ond
meunin
Example The word " Bat JChomonymy word
becae bat can eUAed o hit a hall or but úa a
lyinq mammal olho

c Polusemy :
i _ C Careelk wurd
ci Polysemy meuna- ny Siqoo
t

Cii LE aword or phrcune with cliEfenent buk relcckec

Senme
 Example h e word Bunlk io a Polusemy uword with
the followiaq tfeneak_meaniay
Ci AA Raanuiud instihukioo
T h e buildinq in whith such an iaahhuhio0 io wcated
CiiiD A Sunonum tor relu on-

CL) Synonymu
lhen huo OY Dmore lexica Fenma hak miqht be spelt
clisioc h e he Sume O r Similan meaninq hey cune
called Sunonymy
Example:(Joh,Occupaton),CLaye Biq) CStop, Hailt

lcs) Antoau.my4
CD Aabany.my Semcuahc 9nelakio.oahip hat hulds
between worclo hut expres Opposite mecuning

C2 Example The worcd Dearls u_an Cinhonymy oE iqht

Cand white o_ncuntonymy oF bladk.

C6 HuPernomuy
ia word u i h a more Cenencl Senpe
C A_ Hypernomy
C hypennomy or CcLn_ cund a
The uword auomobile
ud.
whenein_one lexical Fenm i C
C2) I n h e relcuiomhip
Ccoaktueak of Some cungen entihy.
Example lheel C meronym of ' Aubmobile-
 C ) Meroo mu-
meronomy neen h a Neluionship uwheneio- One lexical
erm io Coastihueak _of Some Lanyenentihy
E xample Wheel is c menonym of AutomobUe

C-3 wite_a Short Note on Wordo et

wlordner
COlordNet ia he lexicaldatabame ieclictionany for
he Enqlish
J
LCunquage
C2) lordNet_coaist of hreedatabuwe
One hr Noun One hr Nerh and one for buth cadjecives
Cund cacuerbs
c3) lordNet ist cl enaea of word eudh senne
belonqiag a clifferent sumet A Se0ne entnes

Comist of set of SynunymaCand aqlcu

C WordNet_u Freely cund publiy Oaileble hor clownlous
Thaa
h a been developedhor othen lanquaqe uke Eurone
and Hindi lordnet
CSHupernymChaia for Riven
River _a anqe aatuna stream o waken Clargen han
Creelc
Strecam, uwuten coune - a natuncl bo.cy of Munning

waten louin or unden he ecmh.

b u d y o f water he part of he earhs Lce Couenecd

wih wuten
 *ApplicatiooaOF WordNet
C lord Senae Discumbicquahoa
WordNet_combines eatunes c a numben O f ohen
reounces Commonl n clinumbiquahioa uworl
Thi mulkesit he best knowa cund most _uilized
cesounce or word sene cliscumbiquukioa

c2) Auhomahc uenyExpansioa

WordNet Semaahic relahiono cun be uecd o expand
cdocument i nut
So Hhat he Search_hor _a
e
uen es
ne

of query term, but

coohned b he paternmatchinq

COverS SunonYmA

Cinc Ccate.qoizahoá :
C3) Document Structuins
informuion extracked From uwordnet cund
he
TheSemunhc
hawe been
wordnet_Concephuel representabion of knowledge

Ued or texk cakeqoizuhona

C Document Summaizahon
OundUnelulCppliccuhon
in Fext Summan-
lordNet uw
zaion
Cund Elhaucdad
Theopproach presented by 6arzilay
uilizea iatormauhoo . m wordnet to Compleke lexicc
chain
C4 hat_cdoyou mecun by lord Sense Disambiquittes
Discu theknowledqe bed_cLpproach

Ana
Ana Word Seme Disambiquihies [SD]
O alSDu a lmportcunk methud of NLP, uainq it he
mecaninqof u wurd Can be cletenminec
2 \alSD Solues he ambiquit uwhen hut CLntes whiie
cleker.miniaqhe meaainq of he Same worcd, when it wed
ia ciffeneat situatio.o
C3 SD Can be uAeduwih \exiwqraphs Much of Hhe
J
moclern lexicoqraphy in Corpus bcuec
J
C SD ia_lexicocruphu Ccan provde Siqnihcant textual
indicabrs.
CS)WlSD Can cilso be unec in text_mininq_Cund Lamrmaioa-
Extrachoa tauaka .
TE Co be ned or he Correctabelliaq of worclo,
becaune h e om.cuun CLim otS D io o undenstand he
meaninq of a word Caccurcutelu in
a parilan Seatence

Knouwledqe Buned Approcuch

1 Superised WSD
Superised approuchesrely on a Senae kaqqed traiain-
Corpun r clisambiquahi on
I CLn
Cppliccahion of he ML_Cappruach or Creaiaq
C classihen
 Z UaSuperuised Approcich (wsD)
osuperNised approu ches mcalce WeOE rOw_Or

untabelled text_Corpora or ruininqCunc cnnutated

clataJor evaluaioo
Comidened C a

CD UAupenuisec wsD Alqorithm CCLn be

Ruwteniaq taak
he penormance
ctii As in partof Speech taqginq onouen side cu
wull
he unoupenuised cloonthmin
Companed o Supenuised culqonthma

Di chioncury
Discu bcweeapproach in NLP

Am * Leslk Alcqorth.m
in banec on Hhe auumphion Hhat
The lesk cdqonthm
J
oeiqhbounhuud will tend b shane
uorcdo in aqisen
a Commoo topic
manaen he leslk Alqon.m_io o
C2)In_a Simplihted J
c_CLn_Cumbiquous word
Compane heclidhonay- cdelaition
uith ternma Cwotained ia ita neiqhhounhood.
c3)The Correck meanin.q C in _ qiven
ecch_ward

cootext iclekan mined by noiaq- claun Hhe denne hut

ovenlap he most betuoeen ib dlicionanycleinihion 3
he qiven Coatext.

CLO Leak_alqonthm CLnOutpenorm he oniqinal

J
cdelinikon
h e alqonhm, boh in tenm c preiaion cun

eliienuy-
cS) Lesledenouledqed hisproble.m cund menhionecd ha
Hhin clqonthm
mu il h_conectls clisambiquate C

numben CumbiquoUAuA_worclo o worddo w-ocLUned

between he Context__and denne ddeinitioa
uword.
C6) One w o hundle Hhi adk
cn evidence prablem
lobserued by leok ún o expund the deiaition by-
lincludin wrda relcated to, bub aut cppeaniny in
indiuidu denae clelinihono

Example
Ln he sentence T went to bank ho_cleposit cLah, Hhe
uaorcdlep.osik_cpoecars inh e Coatext of Cambicuuu worc
bansbuhe ceiaiion banlk ia Amencun Herikaqe
IDicionany cloes aut coataio deposik.
\me can_expand he Sene deinition c hanle by
induding cleposit