UNIT:2 basyath,

clAsSMAte
Date
Page
Narrdam.

Biy Data Frameworks : Hodooh Apache Shonk.

*Hodoob
Jhe Hadoss eCsAystem ia a duite t tooland famewsk
duignud fordubilbutd stoage and jisuing big
coll
data, tbicaly aces mehy machinesi a
eectiu andl scalabe ulay
Oozie Chukua Flume Zokeper Dala Mamagement
THiue Pg| Mahut Avso Sqoop Data Access

MabRe uce yARN H DateProcessing

HDES HBASE *pata StoHage
Hadacp EBsskn
XCoHe Comborenki Heliable Soyage
1. HDES(HadoopDübibukd file A
Syskem): scahble
2.Mob Redue i Abusguamminq madel yor hecesing lange dát
and Reduce.
S d wies thostebi : Map
HeAGUNCe management
3. YARN Yet Anothe Resounce Negotiaton)#AhAecs aeSH Cluste
and dcheduling
|ayn t
4.He: o m Austcn
Adae , thdt þrouide SQl- Queeing
Qnalyzing laye dahuet uing hih-lial lang.
Ablafoem
5.Pp: for
Cidabase forhandlng teal- tine dáta acces
6. HBose No SQL ice
A distibuted coondinaton
7|Zeekeaper : ingestien tool.
8.Flume and Sgoop: Data
9. Ogie tt wsK;ews bchedilen for managing Hadssp jeht
data lnpoHation
 classMate
Date
Page

Diadantges
Eis tault toleanel
-i) Hrly aualloble
i) Stabilit iuus:
Secwit (oncens
y Cemblexityy lakneys limiled
autpot for Skavi~kored data
tAboche Sbauk bewCl, Uni<id analic
Abache Shank s an Shon- andl
buscesing- dt sustot beh bafh
engine fen big data in 009 in UC Gerkely
steah bbcesing Shank" stanled
RD Lab and in boit became Gsen seue wnder BsDliceñce
XOmbenints tf Aache Shask: SPark CO1e:
SPank
SPak SQL_Shearing Henenal Executve engne
Fundamental lnit?
SPork CDAe
R "In-memory Combutaien
APIScala 'ython Jawa Baste /o funckonaliby.
dt ç genenal exeuhon engine for
D Abache Spak oHe : API that defines 'RoDe
spark plakom Home "to knd
his tompenent sheratu uith many
i SbakSÑL:Structuneod data.

deta in Aeal time

MEb:Machine leanning hbrany includes a uide hanR g
ML algerithm cellabdtie tering, clueing, kegseition,
VGabhilibary fo. manisulttng gsash gdalathis and
Zonjlement tomptations lniti
 classate
Date
Page

o Key Feakres 3-Th.membry Bousirg loilid Batom (Bach +ReT),

fault Tolerance séalability , Ease f Use
" Chllngs: laemcy dusue ,Complaty n Managing starte. s
Raseuwvté Managemtrnt, Tning Peifomane
Featunes Hadocþ Apache Spok
Procesin Batch þYOCUSSing Batch amd RealHime
Speed Slower de to cigk y/o fasten uith in - memory
Comblee ues- Fiendly
|Usa CasesDala Stoxage and ETL Aduanced Arnatytis
Real-Time Big Data ProcaAsing: Apche Stom and Flink
* bache Stom
Abache Stom isa distuibutied teattime conbtatin
Byetm nw for ia sipliity, scalabiliby arid high
t2 paxtiulany suitd for piecessing nbeunded data
6beam
* Achitcwe
Worke
Suporuise Hhlorker Node
Zockeaporl
Nimbus
/Master Node)Zookecbes SapeuusOLhorker
z0okecpodLSubeusOr HNorker orke
Process
Cuter Executor Task
C0ordinaton
 cAssHate
Date
Page

DNimbus : he mastor node asbonsible fo distibuting cedes

Qsigning tasks,ahd monibining kmputaton
i) Suboruisoil : hdorke hades hatCnecutt trtt awored by habes
i)ZockecbeLiA c99ndnatin sonice enawning cluste's Hetabili,
TÖpolegy: A'tapolgy" in Stoxm Hiþuuentf adata eo graph
"Sheuts: Data SoniOs emitting streama e tuplu
"Belts : PrBCeSing Unib hat tonáume sheamy and emit
tramstoned data

* ue Cas Real-time analytes sFaudl detectin in finacial

* key Feotumes -Real- Bme Procesing , Faullslenance,

Scabbtlh ntqyakon u~th ether syskmds
*Challngu Seplanning kuue for beginns, Hiphen
HUselrce censumplien Depencncy on Koskesbe.
*Ahache link
bache Flink isan ben -Bewce, distributod shea
-beoesng uameuek hat BRcels at beh bateh and
t mepessing. FAink þhocesLS data es beundled
eal-Unbounded sheams
+FLlink ensoLs Hel ability uth theckjsintng ond teveuy
*Flink Architeckure
Eeink's anchikckne lenslst fa job Manager and
 clAsSHAte
Date
Page

Job Manaqu : Jhe masb nede maupensible jo managiny

nd
he Crecttien st jobs,inchuding HessUnce állocaten
ob scheduling tak in bavalel
ask ManagcH:hlorken node that xete he
handls he achual þrscasing y data Task
- Managet
Task Task
FLink Cllent Progann Fnk Master as Task)
Jeb 4 Data Streamy
Submt Spakhn
Monsgor
Schedulen TOAK Manage
Buil der Resorce
Cant Checkbant Task
Cooxdiale

*Cone featvos :-Unitied þrocessing , Event-time Brocessing

Statchul Processing fault Toleiance ,Ferible APIA
Lou latencyScalbilty s Higl- hreughput.
*Challengsi- Ressuxc Management,Stat Managemant
Atache Stoum Apache Elint
XAspect Tybe Pure Sheam Procesing niied bateh ¢RT.
J.Pocening
2- Lakny LowLakncy, event-oiea lba low laercy.
3" fault Toleronca Basic, task restts Advanced,Checkpocnts
4 API Vsabilt Louwwrleu ig-luel-APTS
5.Statemer Limiteduitheut Natiue, obust
-

Trldent
6-Event-tne Limike Comprehonsie
walemaks.
uith
|Suyport

Easien to sthuh More comex but

featune -ich
 alassmate
Date
Page

Big Data Visuligatien TGels: Tableau , Pouer BI , Zebpelin

Tableau
ableaw i a wldoly-lsed data isualigatin tosl
Senoned fe ia htinactiue dashberna and ese of luse'
* TableaL Sewe AHchitechue
Data sewer DatoWarehouse Daia fYanti Ates tcubes
Data Conn ectorsFos Dalo Ergine SeL onnecloy MDX Conn ecta
Main Combonents Data SowerVz0L Sewer |ABblicaWon Sewe
Gateiuay Graleuay/Load Balancor
clionti Desktes Mobile Brauser

*| KeyFeatuxesiPertomance
dnleqraBienPrBesingsdntenactine
wuith Biq Data Platonå, Dashboand
tlgh
Ease sf lse, Adianced Arnalyis 7Scalhbilty,alabonatens
¢ Sharing"
Lae Cases ;-Busines dntaltignetRitostinqsCuomes Analyia,
Heathcane Datk Anaysi, IdT Data Monifong.
Chalenge:-Cost, leaning Cunve ,Jesendeney ot Data Prepavahon
Pouon BI
data viaualizatien toel that inabes Aen to ©nalyge big
data and reate tinsiphtul, intratiue aashbedid an
* Key FeakoresiWide Range 8 Data nectiity, saalaliltby
RT Misualgatien, Avanced Aalyts andAI Tntepaten
"Seamlet Colaboahon. rle- leuel Seciry.
 clasSate
Date
Page

*fowey BI Axchitectwe
Powen BI Seice
Power BI
Data Serce peliverg
4Powes BI Rebort Sewen

" Jhe whele breess at data sourcing to he cneatien sf reponti

Qnd cauhboawds censist af 4basie steps
|313-Rebert
Seuscing Data. 2-Tians7oming the Data
ahd Publisk 4.Creating boshboand.
"Achitecure Componemts
PowewQuery:-tallsus uAer to lonnect distincd info om mulhile
i)l Power Bx Souicoss- At Conneck othor Comporent uth eaihothe
i) Pwer Pyet :- Data modeling techmigue to Creale data models
i) Power View-t CTeates gábhs, mabs , Chant, ete.
RuwerBI Deskteps- Brinq eieuthing on a.Singe laorm.
v) Pover QAA- t use NP to gel he ansuier to yen qey
*Abbl8atiensi- Data Visualizoton and Data Mancgement
"Data Analytes with Inkunal Softsaemasketing. Sysena
"Gustnes eposting Qnd Enhance the
Vieus in fousen BI:- 3types f Wewsand you can
Suuitch betueen he 3 iews wing Nauigttion Pane
)Rebort Vew WData Vieur () Relaionshí,(odel) iad.
*Challengs-forformamce BetMemucks
Limited ustomtgatiens
2Oning owe fo Aduenad Fiahves
 alassmte
Date
Page

Abache Zephelin
Apache Zebbelin is an bben-Sewne web based nptthock
Hhat sushote interactiue'data analys Nismaligatos
and collabonation fo bip data ueitouo JVM
* AchiBecte-JVM- spank Inerpren rOub
zeesetin Sower fSpark SFanKSEA

WebuI Tajo Tntenpreter)

Port zeepelin Engine - JVM
Fink Tnenpheler
zeebelin
Ina Cauanda Tntesprchka
Zeebelin has aclientSeruer archtlecture.
heclint a ueb based itnace where wser tntraut
netbotks he sever manags intbekr and laate
um
Lcemnond also handls becirity, Quthentication
* Kou Faates:Mult-languag susjert, tnlenchveNokeboskL
Viunltgalons," oExenstbilty, Scaakbty
-Dynamic Data
dnkepaton uith &lgDaa Tiaylem,
*Uze lase DataExplonaion and rebrocsng
" Machine Learming lokhoues
"LBg and Terxt Anatytis
Sttaming DataAhalyi
Challeng leavning ure
"Resewc tndensie
"Limited Buit in Vaualgatiens
 clssmste
Date
Page

1- Renl-time. Analytis dnesasing demand fu intant insight

Neels such as Atache kaka , Funk aneused Jhisi aluable
in secto like fnance, heathcare, and e-Commerce.
2- Data ObsQndAuomaon' Data Oþeyatons focUsLS on iinbbinq
Sþaud, gualiy and Teliabilty data analytu pipaliná eea
such asau shache NIfl, Abloo "Efient managoment
tenatiens hdata
fe
séient
Complen ETL bipelenes y fasten
3AI and ML Jntegaken'AdvDnced algonithms enables
bedictie analytics, teal time dedion- making, ete
Mub "
ML kamuoYk lIke TerserFlew, h orch spånk
Redtue analytiu in maklág f Heathcane,Autremussykn
4- Edge Cempuingt-Shiting data þsKLLing cleer to the dhta seue
te educe latbncy and banduidn usages. Jess like , ete.
AWS Cneenga, fzune IoTEdge beis Rr áT analtiu
5.Puiacy- Enhacing hchnelogiu(PET,) -dnenasing feeua bn
secwie data piesing dúe to shingnt hegulaken (GDPR,
CCPA). chnslozia like Bsyft
6.ALgmented tralytiut- levengng AI andHLto Autmate
þredietiue anaytis .
data baesaakieng insight gehonaten,Vhakon
Jesle ike Qik , Tabeau and IBM Analyti
re> utomaled tehoutin and decisbn-maktng in busines
L, Personaliud Aecomnendatbg:
 classMste
Date
Page

* Databasi- Adafabae is an Drganised colltenstdati,

áe that it can be lOsily accud and managed
"SOL(Shrucaed Queuy largunpe) and NasOL (Nof only Sa9
* SQLDatabases
ASQL database ahe knaun asa Hlatonal datobasly
isa systmthat stos and sganigu data into bghly
stuctired tables t teus and coumns
SQL datobaseLi- MysOL, Dracle Daabese, Mineet SQl Sowe

* Key Featnes'- Stucured Data Model, Schema-Debendant,

, Dwable
ACID Cempliance : Atomie, lorsistent, Jebtion
"Standand Quevy longuage: Sql for quenying and manmipulatien.
Seanith "Pwextil and ersatile
*Disoduantages
K /limtotiony i Scakbility limitatins,
Fred Schma, Not Jdeal toy Unstchwed pata

* Use CaAes i- fnanial Syskma- Banking, Qceounting, ete

Enterbsise Applicatien, Data Wareheusing
* BdsQL
Big SoLis an SQL engne Duigned to un SQL Quexis
o bip dota stond tn duthibutd Aytms like' Hacogs.
*Keu Fatwes i
Handling Shrucksed Date SQL Cempatibiltty,
SchemaEntorcement Qnd alicotien,
uith Rlatenal Data Serces.
Sateggtien
 classMate
Date
Page

I) Supþat for Semi-Shuckured and hstucured Data 3

Susbet for Semi-Shuckned Formats, Flexible schema
dntqraton uith Big Data Storgc Systim
) Roblst Queuy opinizaHon (»Unifèd Queny ngine ACID
Compliance MHh Persomance uD bat'secunity
*Limitatiens:- Comblex Selup lonttguraton, limited Flexiti,
,Ressrce tntensine, Vendo Lock-In,.Cost,
Lateneyin Real- Time lue Casese
*se Casesi-Data Wayehausing, EIL Pialines AL ML
Bustrendntlligenca 4 Rejantings IoT#log Amayis.
Biy SQL Erngines -LBM Bigstlsbrogte Bigluexy, Ahache Hive.
* Query Cbtinizalon Technigues in Big SQL inpreue
uay phimipatien in Big[oL is critital to ua.
reseur
borhaiHmance, Heduaingeeutien timet
iCest-Basecl Obinu¡aton(CBO) - To sysems, CPU, memor
t)Join Opimiz?: Jin Order Opimz, farhhoning AuwaHeness
it)Tndex VHllizaHon : Bg SQL Can lse indexe to sheed sgueryxa
querie
latency f subseqentAralysia.
t) Cachingi T6 1edieo)queniy
VPagat Exeuton Ereculton Plan
Featwres SQL NoSQL
i.Data MOdelg Relaional (tables, Yow) Non- relahional (Vaniel)
Schema Fixed Dynamc or Shema-ley
ii.Scaling Veical Hon'zontal
Shong AciD Compliae Evental or Flexible
iv. Consistency
v. Qery lang SÍL Database Speafic API;
Vi. Best For Shuctured data Unstuctured dafa
VIt.|EXamples MysQL Oracle Mongobß, Cawandra
 alssmate
Date
Page

*NoSOL DatabaAes
NaSOL databases ane non- Helational and designed fo
lenibility, scalabitt,anad handlng nstiuchuud
Examþles > MongoD8, Cassanda, Redis, and Couchbase .
*Key Fatwesi
Flexible Data Models : Cam Stove dataas key-vale faiy dlocumendea
Scalability : Horzortal Soaling andEastie sealing.
Big Data Handling "Fault Tolrane High Petomane
Donammic Schena oCat effechie keat Tine licatins
* Limitations:-olack of AcID Tonsactens "limibd uengirg
"ata Redundaney and Sntegiy " limitd uypost for ote
" lack ef buill -in Reportng Tlol " Dshibuted heiiectue
**Use Cassi-Big Data dRT. Analytes Social Mudaf Recoamndt
Enginw IóTPSenso Datl "Commere "CMS
*Types ef NoSCQL Databases
NOSOL
|Key,Value DB Doupent DE Colupn-DE
Redis 'Ceuch DB "Cassanda D8 Neot
Dynano DB "HBQse
MongoD8
1 Keu Vale Stoves : Stores data as key-value Pains
2-Deument Stores: Sores data as JSON- ke douuments.
Stores data in ceumn-Oninteo
3.Columm- Family Sores i tomat for tast fetrieual
4. Grebh DB: Store data s NEde Dnd elatienshs, seful
foe soal metusks and recomnendateng
 clAssMAte
Date
Page

Doument SBores (MorgoDB , Couch DB)

Mongo DB
MongoDB is a leadng cecumont- Oiented No SQL datbase
knatn fo itt bcallbillth and deuclsber -priendy featua
t Uses tsON- ike BsON chjeets to stoe data, alliung
o2 a fentble schema
APlicaton
Diver
QuenyRouter Quey Router
Shad 1 Shod 2 ShandN
Pimary
Primary Phmary
Seconday Seconday Secondy
*Key Faahuresi
Document-iented anad Schema less Database
Storage Fomat:tA0s BSON (Binavy JSoN)as it inlenal data fomat
DocLLment Model : Data is stored as document wtthin Collechio
"Queuy Language: Use a Hich and exible gueny lang totextHehieue -
Tndexing: Sufpot Single jeld, Campounds geoatl t indenes
Scalability: Hoaizonkal scaling, it supfoti shanding
Resicalon : Suspokt TBlia sei to ersune tedendancy
alaiablty
JAe. Cases: Conerd Maragemvd stem, Ref analytieiy
Social medla abblicahons , e-Commentt plt.

CouchDB
Abache CouchDB is an oben-sawre, NosQL doument
datobase thot tocusts on dcalability cay fuse and
data reblicatien t ses a schema-ce, JSON-bad
stoage model and RESTul HTIP APL foN data
intokacen
 clssMste
Date
Page

Docmant
Raplia Document
HTTP Couch DB
REQUEST Engine Reblica +Document
DB2 DOcment
RD83
* Key fëokunes:
Storage format: tstores data ay JsoN document , Sehema- free
Document Model: Similar to MongaDB.
Oueey Language: t uses MabReduce to cHeate uiew.
: ReplicaHon Jt ecces in Multi- master ssliaten
Scalabilit i Support horizonia scaling ttosugh Cstering.
*UBe Cases i- 0Hline- Fist Abbs, Mobile Sync .
equining fst, Hesitle
Noe Mon go DB is beter sutted fe2 Qsslen Conpleo
guevying scalábilty and agpephen Capab
CoLchDB ercels in enuironment that need subpont fer
slicaken s dstrikuted yen and fline capablitet-
Ceumn Family StoHes(Cassandwa , HBase)
*Cassomdya No SQL
Abache Casanda isahighly scalable,dsbibuted muiple
database deuigned to handle bi¡ data acres
tolenance
Souers with high avaiabilithy Qnd faut

Data (ente 1 Data (omte 2

Cluste
 dassMate
Date
Page

Node : is basic Combonent in Cassenda . t s the þlace hehe

ackually data iu stored.
|Data Cente: tii colection et nodes Dc NjNstNgt
Cluster: dtis collechien ot many data Contres
* Keu Feotunes:
Dao Model -Cassonda stores data in tobles, hich ane dehned
by Bimakey s which is sslit into 2baat:- faattion ey amd
*Reicatont Data Cobied at multple hodel. clustorigkey
Conststency leuals: Tinable Consitincy louels
+Queny Lamguage:CQL (Cassandra Gueny lamguage).
*Use Case:-Time-Series cha, euent legging, tecomdlengne
tHBase column-0sientd , dstuibubd databae
Abache HBaueis
btlt on teb st Hhdseh t subbent teal-time Head!
hlitt aces to lang dataset
CAent HMaste
1RegionSerwer Z0okeeber
Region Sewer4
Regien Region
Regien
HDFS

canamda
Data Models Similar to into regiong.
Regions and Rogibn Sewer : Dta in HBase bbaationed
Data Acce fouides ou (ateney rad-wnite Obenatong
Consisteney : Pouide Shong- Consstency at tsw-leuel
tnkeasaheA ILih Hadoep. "Tnteqyaked uel wuit h Hadsep
 clssmste
Date
Page

* Adantae alabiliky tighfoyfomane

"Aexible data model faut Tele ance
* Dindvantagzs-Complexhy liniled Quy tanguage
cHons
NO subpt or tana
* UAc Cas:- «RTo analyties olog- Rocastng aStoring Semi
stuchoed Umstuctued datr
Ker-Volue Stores (Redis, Dynomo D8

Redlis
Redis (Remste Dictonay Sexver) ua high-Porformamce,
in-memny NoSQL dadbase buimarily used for caching,
yeal-time anayties,
UMain
and mesage brekoing
Reicated
Instance
Rebliakon Read Clet
Read Cint
Kecp data in memors fo fast acces lCan bersist dta
en dik Ghenally)
DataTpe Suppouhed- String, hashes, list, Set,Sorted Sek
*Advantngl:- Extrerny Fast, Subpont atomic obenaion on ds.
Digodu - Memory band, Not auitable fo Canjlex Quwying
XDunamo DB bcalable
Aullymanaged, highly aniable, and
Ko-alueand deuhent atabase seuie bueuided
b Aws.
toes data en SsD-backend torage uith ephonal
in-mernly Laking (DAX)
 classMste
Date
Page
Pimoy key
Amazon
DAX
Dynamo Table
Clnt AWS Amazon Pk SK AHihety
Cloud Shell Elasfc Cache Dynamo
OB

pkt Pavtiionkey, Sk: Sont key

* Advantoqes:- -Fullymanaged sexuia wih auto -scaling
"Flexible Schema lobal tables for Cren- reion Yejltie
* Dizodvantages'- Exþemsie s Limited quexing.
Featres Redlis Dynamo DB
In -memory key-Value t Persistnt key-Value StreL
iü) PertormamceUlba- leo latency t tigh Pertomene.
i)Persestence ObHonat (Snepshot ) Persistene by deign
|Mamual uuith Radis Clt Automatic Scaling
)Tnkgyaton Oben- Sowrce, Mult Ca Natiue to Aws Besytn
vi) V8e Cases Caching, mesdging, High -availabliy gfs.
Real time andlyie e-commehce:
Vi) Data Type String, bashes, list, Scalay(Shang, Numbos, Boc!
Subported Ses Sorted Ses Doment( JsoN-Like obyect)

GiYabh Database (Ne 4)

Neo4 l a grabh database designd for strngdata.

änd analsing hiphly onncted
9uniying jsFriendof ikes DN
serves
Cutine'
Locaien:
sevs Mohali
name: LOcated n
(Sush
 clsSMate
Date
Page

Nodes: Rebyesent enites Buch as beoble, bustnoses , ek

Edoe (or Relatierships): Connecs nodes and ilustales hsw entitu
aHe Helated
PrcpedHes: Prouide addttienal intomatien about nodes
and Helationshibs '

* KeyFeatuxes: Native Gtraph Stoage and Poamirg

"Cyphen Query Lamguage Neo; use Cypher
PCaph Algoihina: FageRamk , Shartestfath , lommwnity
Daechon , Cemtraliby .
ACID (ompliance Tndexing High Penfomance
Scalabilily isualijalon Tnkyaitan uith tchnalga
* se Cases' Social Nehuok -Recommendaien Systm
Susly Chain Management , Netuork and IT Gzoratiens
Memouy-nlnsisescaling
* DiAadvantyei:-Leanning lomplx
ue, Not Idedl for sime Lokuj

*AI in Ba Data
Jhe integiaon G¢ AI and ML into the domain &t Bio Data
enhanis dlata puocesng, dlecision- making, fuomatin
+AI and ML þouide the'conputahonal peuer Qnd algo
þaedictions
to analyeBip Data, neeuer fattekns f make
One y he nmast bignicant abpliahien i NLP.
*NLPNawal longuage PDCM0n)
NLPisa branch f AI that nables mmachnes te
endeutmd, inlepet, andqemesat human languge
 clssmate
Date
Page

*NP Technigues
J Text Poesing and Preboceslng in NLPi
Tökemization: Diuidirg text into smaller Unit, eq hlard or Semloncs.
Stemming t lemmatizaton : Reducing word to hein bau fom(kk
Stopuoxdkemoval: KRernouing (omen uords (like'ard", "s"
Text nonalizakon: Stardamdizing text, punchuahong
2- Sentinent and EmoHon Analyso in NLPtmgoaHig: speling
Emotion Delecion : Talentikying amd (alkgosizing embton expyesed in text
Cpiniern Nintig: Analyzing'pinions to undenstand þuttie
publie sentinent
5.Spezch Rosng: Kpch Recogniton, Text-to-Speuch (Trs).
4- D'aloguUe Syste' Chatbots anil VirtualAsiutants
Predtckve Analtte : tt ses AIAML tachnigues to forcat
tutne euteme. Data Collecion, DaBa freproceking,
Featre Engneesing Statstical tralysit ,MLmodes.
Tianahon , TextSummani
5. Lamouageenena Son : Machine
-Zation , Tect G1eneHon

* key Athlicaten g NLPirn Bio Data

D Tet rnasts anad Sentime¯t Onalysis.
2
ntomatn Retieual and Seanck Engines o
3) Chatbots and viytual Asistant
Atomakd Summariation
9Speech-to- Tent and weice ntertaues
*Challengei- Scabbilit,, Diueraih GfgLanguage, Dah Quali,
*Fitoe Directonsi
D Mullingual Gres- languagc NLP.
3Real Jine NLP
9 NLP n Healhcare and Life Sciences .
Explainalble NP mnodels.
 alsmata
Dafe
Psge

IBM Natson with.

IBM Natsen is an entvbsiue-ghade AI Seruices
anolytis
and
teal tor aata þuocesing, automaton,Jesfavdy,a
Chapion
9uie sheu where itl outpesfomed hunan
Watson Hnduane: 10 1efigevaon siged Sylem.
8core,4SMT Hreds
þyocesors ’
92 POWER 750System ; 4Poert Evidence
Soeree
Supolng Deeb
tantiat ehieval Scoing
Jams generaion
/Question SeancK

TH9botfhess Sott HghoihesiSnteiFnal

Quey l4eneraton Lilloung and manging
Ouesorlecompo2to
Anauis
Scorin Trained Ranking

HybctBhari SeHt HspelhciS

generato 19i14eringl idonce Scoring

inalyycs
Seanch ixt in mullle lang fu inise.
LNes Cognthe Voice aistants
2 Conyesálional AT: Chatbots dndConvents &peach-to-text f ua-lea
3-Sech Recogrthon emexahion
4.A1- weed AutomaHon : Reducts manual etorts
5Tndusty-Specife Solutons
hatson for Oncblegy hels in toeatment blanning.
"Heathcane: Hisk asesment, busonalized bankina
Frnance: Faud delechon ,
stomer Mecommed', Inwentry OpHnization
Retail : Porsonalied Leaning.
Legal t Educaen: Contract analyi,
*Challenges i- Data Puivacy Cempliance
"Technical Expertie.
Cetly lstoiFatin Niede.
 alssmate
Date
Page

*ateon Seuices
Key hlatkon 6erucs aHe
Aoson DeovKY: A1-foweNed beanch and analyts brgin
ddealtaA exbatting Inatghts fyom shuchuned 4 Unluchuttd
data useflis bedmcane ,legal f Customer seuics:
key FeatweL b> NLP, Document clasihiatbn QlA,
mocdels.
Pre-haned nodlls
Aanced Search Catabilitier
2Watsen Studlo: lonþsehensiue enuinenment foe duýning
raining, and desleying ML modls uakáing
Peyect fo dah scentikl1, analyst and deueloteuele.
npnane , retail,
n custon pedthie models Lyeycle , Brigamming lang Sufert
key Feaheres t> End-to- End AI rteton
GutoAI Collaboratien Fahres , Coud
3.Waton Assisant Conuersatenal AI platssm for bildng
ntelligent litual agnt ancl chatbat.
imþroving us¿n laperitne Qnd autmatng
deal fer velail ftelecomunicl.
ustoner beruihin baikg, TempatU, Conkxd Manage,
Key Featoes:> NLU,Pre buil!
Backenddnegyatn. Liu Bqent Escalatiens
in Bia Data Decision -Mating
*Naten struclsed fUnsbikadda
doka tfa
* Role :>"Processld vast like a human
"llsesNLPfML to Hestond deliuers kIinsight.
ddentfispattern,tends, Qnd

Tnkqsahon wih OhetSkns Openahönal

 adassmute
Date
Page

bhamcs acssbilthy sy 9I tole besos ndustu

pimiye Bpenain by pdictingdLhon t Bea
Tntcgrakon fhatson uith Big Dah totle
Bntegsatng IBM hlatsen with Bigbata tusls can huls
Bgahigatibns leuersge c0gnitiue'cemputing to eatiuat
ihsiátta , autonat pebses and enhanb cleiion- akig
*Big Data Toels t r Arteqation
Hadoop Ecesysemn HDES, Hive and Prg
Apache Spatk
No SgL Dataases MongoD8, Cassandra
Relatonal patabaes - Rstgre SQL s My soL
Data isuaizaten Toels 3- Tableau, Pouon BI
ETL Tols TalendyInformatea
Data lakes- Amagon S3 zwe Data Lake.

Snteqyalon Methods
+RESTl APTs "IBM hlatson Studio "lutam Connechers
IBM Cloud fak tor DaBa. "Evend Sbeaming- kafea
Use Case-Data Ennichment, bedietie nalytids
Sentimet rasis, Chatbot and Vahual Astat.
Healhcar Analyte
*alalson Corobonends Wason Assistant, Watson DjsCovexy
Watson Skudio, Walson NZU, Wason Vual Rec9nihon,
Text- to-Sßeech Sheech- to- Teot, ete
 Secnity anPrluay
d Data )
loneaithtegratin Data D
*Challenges:
koperitnes:
Jmprovedcustomer
SeruLies Lghod
Seruices - batson
otson baed leud thaughScalabilty
tine-
to- Jastn
utomaHon. making decision 4Enhaced
Anteguatien: Bemfit *
t. habenAsitant
delie
AI do and MengoDB
intenactien
data
in Customer StMorgoDB:
one Wasont m)
nstuchued
data amoundt
of Vast Store Hadoop 1)Natson+
i
analyss cogniiue fohlatson
r amd
dishibuted
daa for Spvk spak :+Apache )ason
Page
Date
classmate