2015 11-4-v14-ickm2015-keynote

59
Dr. Dickson Lukose Artificial Intelligence Lab. 6 th November 2015 Application of Semantic Technology in Knowledge Management 11 th Interna*onal Conference on Knowledge Management November 46 Osaka, Japan

Transcript of 2015 11-4-v14-ickm2015-keynote

Page 1: 2015 11-4-v14-ickm2015-keynote

Dr. Dickson Lukose Artificial Intelligence Lab. 6th November 2015

Application of Semantic Technology in Knowledge Management

11th  Interna*onal  Conference  on  Knowledge  Management  November  4-­‐6  Osaka,  Japan    

Page 2: 2015 11-4-v14-ickm2015-keynote

Contents

•  Types of Knowledge Management •  Data, Information & Knowledge Challenges •  Semantic Technology Platform

– Text Understanding – Natural Language Query – Data Harmonization

•  Knowledge Management Portals •  Social Media Intelligence •  Social Network Intelligence •  Conclusions

2  © 2015 MIMOS Berhad. All Rights Reserved.

Page 3: 2015 11-4-v14-ickm2015-keynote

People & process-oriented KM initiatives •  Cultural and Readiness

Assessment •  Formulation of a KM Strategy,

Framework & Strategic Planning •  Knowledge Audit & Knowledge

Management Audit •  Change Management •  KM Assessment including the

definition of metrics & reporting of Intellectual Capital (IC)

•  Community of Practices / Special Interest Groups (SIG)

Technology-oriented KM deployments •  Search Engine configuration,

testing & deployment •  Taxonomy development,

maintenance & governance •  Collaboration System •  Enterprise Portal •  Electronic Document Management

System (EDMS) •  Knowledge / Information

Repositories •  Content Management System

(CMS) & Applications (CMA) •  E-Learning •  Intelligent System •  Blogging / Weblogs / RSS

Readers / Wikis

3  

Type of Knowledge Management Projects Technology-oriented KM deployments •  Search Engine configuration,

testing & deployment •  Taxonomy development,

maintenance & governance •  Collaboration System •  Enterprise Portal •  Electronic Document Management

System (EDMS) •  Knowledge / Information

Repositories •  Content Management System

(CMS) & Applications (CMA) •  E-Learning •  Intelligent System •  Blogging / Weblogs / RSS

Readers / Wikis

Lessons  learnt  from  nearly  two  hundred  cases  of  KM  journeys  by  Hong  Kong  and  Asian  Enterprises  Prof.  Eric  Tsui,  Hong  Kong  PolyU  

© 2015 MIMOS Berhad. All Rights Reserved.

Page 4: 2015 11-4-v14-ickm2015-keynote

World of Data

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

4  

Structured

Unstructured

Page 5: 2015 11-4-v14-ickm2015-keynote

Knowledge Audit

2015  MIMOS  Berhad.  All  Rights  Reserved.  

5  

Enterprise  Data  Linked  Open  Data  

Sensor  Web  

ACTIONABLE  KNOWLEDGE  

ASSETS  

Structured,  Semi-­‐Structured  &  Unstructured  

Unstructured   Structured  &  Semi-­‐Structured  

Page 6: 2015 11-4-v14-ickm2015-keynote

Ontology

Technical  WriTng  

Term  Databanks  

Machine  TranslaTon  

Human  TranslaTon  

InformaTonRetrieval  

Knowledge  Engineering  

Consumer  InformaTon  

R&D  

StandardisaTon  

Nomenclature  

Terminology

AGROVOC

MYGMO  

PADIPEDIA  

AGRIS  

HERBAL  MEDICINE  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

6  

SNOMED-CT STW

Page 7: 2015 11-4-v14-ickm2015-keynote

Knowledge Portal Challenges

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

7  

Knowledge  Portal  

Natural  Interface  

High  Density  VisualizaTon  

IntegraTon  and  Insight  into  Social  Media  

CoP  Network  AnalyTcs  

CollaboraTve  Problem  Solving  

NOT  What  is  the  Content,  BUT,  What  is  

in  the  Content  

Making  Sense  of  Social  Big  

Data  

Big  Data  Challenges:  Velocity  Diversity  Volume  

Seamless  Integra*on  of  

SME  in  Problem  Solving  

Page 8: 2015 11-4-v14-ickm2015-keynote

Semantic Technology Platform

8 © 2015 MIMOS Berhad. All Rights Reserved.

Page 9: 2015 11-4-v14-ickm2015-keynote

TEXT UNDERSTANDING

Page 10: 2015 11-4-v14-ickm2015-keynote

What is Text Understanding?

Conceptualiza*on  

John  is  going  to  the  bank  by  bus  

Person  

Human  

Animate  

Financial  insTtuTon  Road  vehicle  

Agent                          Des*na*on  using  Instrument    Inanimate  

Knowledge  Graph  

go  

© 2015 MIMOS Berhad. All Rights Reserved.

English Text:

10

male-person: “John” go: * bank: *

bus: *

agnt dest

inst

Page 11: 2015 11-4-v14-ickm2015-keynote

What is Text Understanding?

Penduduk  mendapat  bantuan  daripada  kerajaan  

Human  

Animate  

resource  organizaTon  

Inanimate  

Knowledge  Graph  

mendapat  

Malay Text:

11

Agent                                                  Resource  originaTng    Source    

© 2015 MIMOS Berhad. All Rights Reserved.

Conceptualiza*on  

Page 12: 2015 11-4-v14-ickm2015-keynote

大卫 在 图书馆 等了 3小时。

What is Text Understanding?

Person  

Human  

Animate  

Library  

wait  

12

3  hours  

Mandarin Text:

© 2015 MIMOS Berhad. All Rights Reserved.

Knowledge  Graph  

Agent                          Loca*on    for  Time    LocaTon  

Conceptualiza*on  

Page 13: 2015 11-4-v14-ickm2015-keynote

English NLP

Sentence

Knowledge Graph

Syntax Structure

13 © 2015 MIMOS Berhad. All Rights Reserved.

Page 14: 2015 11-4-v14-ickm2015-keynote

Malay NLP Sentence: Meningkatkan harga barang dan minyak kerana inflasi negara.

Annotated sentence: Meningkatkan_VB harga_NN barang_NN dan_CC minyak_NN kerana_CC inflasi_NN negara_NN

Knowledge Graph:

14 © 2015 MIMOS Berhad. All Rights Reserved.

Page 15: 2015 11-4-v14-ickm2015-keynote

Mandarin NLP

15

Text

Segmented Text

Part-of-Speech Results

© 2015 MIMOS Berhad. All Rights Reserved.

Page 16: 2015 11-4-v14-ickm2015-keynote

Mandarin NLP

16

Text

Segmented Text

Entity Recognition

Results

© 2015 MIMOS Berhad. All Rights Reserved.

Page 17: 2015 11-4-v14-ickm2015-keynote

Mi-NLP Demo

17  © 2015 MIMOS Berhad. All Rights Reserved.

Page 18: 2015 11-4-v14-ickm2015-keynote

NATURAL LANGUAGE QUERY

Page 19: 2015 11-4-v14-ickm2015-keynote

Traditional Search

Too  many  results  

Pa]ern  matching    Pa]ern  matching    

19 © 2015 MIMOS Berhad. All Rights Reserved.

Page 20: 2015 11-4-v14-ickm2015-keynote

Traditional Search

20 © 2015 MIMOS Berhad. All Rights Reserved.

Page 21: 2015 11-4-v14-ickm2015-keynote

Knowledge Base Preparation

Mi-­‐CLIP  

Mi-­‐HARVESTER  

Mi-­‐KRAKEN  

Mi-­‐NLP  

KNOWLEDGE  BASE  

World  Wide  Web  

Linked  Open  Data  

Enterprise  Database  21

© 2015 MIMOS Berhad. All Rights Reserved.

Page 22: 2015 11-4-v14-ickm2015-keynote

Natural Language Query Interface

22 © 2015 MIMOS Berhad. All Rights Reserved.

Page 23: 2015 11-4-v14-ickm2015-keynote

Semantic Query

23  © 2015 MIMOS Berhad. All Rights Reserved.

Page 24: 2015 11-4-v14-ickm2015-keynote

Visual Semantic Query

©  2014  MIMOS  Berhad.  All  Rights  Reserved.  

24  

Page 25: 2015 11-4-v14-ickm2015-keynote

DATA HARMONIZATION

Page 26: 2015 11-4-v14-ickm2015-keynote

SNOMED CT terminology

311,000+  concepts  

Hypertensive  complicaTon  

[SCT_449759005]  

Kidney  disease  [SCT_90708001]  

NeoplasTc  disease  [SCT_55342001]  

Hypertensive  renal  disease  

[SCT_38481006]  

Neoplasm  of  kidney  

[SCT_126880001]  

Renal  impairment  [SCT_236423003]  

Chronic  renal  impairment  

[SCT_236425005]  

Kidney  disease  

Concept  ID:  90708001  

Preferred  term:  Kidney  disease  

Synonym(s):  -­‐  Renal  Disease  -­‐  Nephrosis  -­‐  Disorder  of  kidney  -­‐  Nephropathy  -­‐  Disease  of  kidney  -­‐  Renal  disorder  

26  4,060,716+  triples  1,360,000+  relaTonships  

           Disorder  of  abdomen    ∧        Kidney  finding  ∧        Disorder  of  body  cavity  ∧        Disorder  of  kidney  and/or  ureter  ∧        ∃Finding_site.Kidney  structure  

Logical  formula:  What can SNOMED CT be used for?

© 2015 MIMOS Berhad. All Rights Reserved.

Page 27: 2015 11-4-v14-ickm2015-keynote

Data  HarmonizaTon  Process  

HIS1  

HIS2  

HIS3  

DB2  

DB3  

DB1  

Schema   Tables  

Tables  Schema  

Schema   Tables  

IT  Staff  (Consolidate  the  data)  

Subject  Ma]er  Expert  (Interpret  the  data)  

-­‐  Time  Consuming  -­‐  Limited  Resources  -­‐  Skills-­‐dependent    -­‐  Prone  to  errors  

27  

q Automate  processes  q Enable  seman*cs  q  Improve  repor*ng  accuracy  

SNOMED  CT  Big  Data:  Freetext  Challenge  

© 2015 MIMOS Berhad. All Rights Reserved.

Page 28: 2015 11-4-v14-ickm2015-keynote

ID   Pa*ent  Name   Gender   Symptoms   Diagnosis  

1   xx   M   FaTgue,  Chest  pain   Heart  failure  

2   yy   F   Breathlessness   Asthma  

ID   Nama  Pesakit   Jan*na   Tanda-­‐tanda   Diagnosis  

1   aa   L   LeTh,  Sakit  dada   Lemah  jantung  

2   bb   P   Sesak  nafas   Penyakit  lelah  

ID   Pat.Name   Sex   Disorder   Diagnosis  

1   kk   Male   Tiredness,  Pain  in  chest   Cardiac  failure  

2   ll   Female   Dyspnea   BHR  

HIS1  

HIS2  

HIS3   SNOMED  CT  (Refsets)  

Challenge  of  freetext  

[371484003]  [371484003]  [371484003]  

[346741003]  [346741003]  [346741003]  

[263495000]  [263495000]  [263495000]  

[243814003]  [243814003]  [243814003]  

[439401001]  [439401001]  [439401001]  

[276179002]  [276179002]  [276179002]  

[84229001]  [29857009]  [84229001]  [29857009]  [84229001]  [29857009]  

[84114007]  [84114007]  [84114007]  [267036007]  [267036007]  [267036007]  

28  

Query:  how  many  cases  of  Dyspnea?  

1   3  

[195967001]  [195967001]  [195967001]  

© 2015 MIMOS Berhad. All Rights Reserved.

Page 29: 2015 11-4-v14-ickm2015-keynote

Mi-Harmony Demo

29  © 2015 MIMOS Berhad. All Rights Reserved.

Page 30: 2015 11-4-v14-ickm2015-keynote

Question Answering System

Mi-CLIPTM

Text Understanding

SQ VSQ

Knowledge Graph

Texts

Question

Mi-Reasoner Answer

English NLP

Malay NLP

30

Mandarin NLP

© 2015 MIMOS Berhad. All Rights Reserved.

Page 31: 2015 11-4-v14-ickm2015-keynote

KNOWLEDGE MANAGEMENT PORTALS

Page 32: 2015 11-4-v14-ickm2015-keynote

Knowledge Audit

2015  MIMOS  Berhad.  All  Rights  Reserved.  

32  

Enterprise  Data  Linked  Open  Data  

Sensor  Web  

ACTIONABLE  KNOWLEDGE  

ASSETS  

Structured,  Semi-­‐Structured  &  Unstructured  

Unstructured   Structured  &  Semi-­‐Structured  

Page 33: 2015 11-4-v14-ickm2015-keynote

Ontology

Technical  WriTng  

Term  Databanks  

Machine  TranslaTon  

Human  TranslaTon  

InformaTonRetrieval  

Knowledge  Engineering  

Consumer  InformaTon  

R&D  

StandardisaTon  

Nomenclature  

Terminology

AGROVOC

MYGMO  

PADIPEDIA  

AGRIS  

HERBAL  MEDICINE  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

33  

SNOMED-CT STW

Page 34: 2015 11-4-v14-ickm2015-keynote

PADIPEDIA KNOWLEDGE REPOSITORY

PADDY  RESEARCH  

GROWTH  

STAGE   SEEDLING   TILLERING   BOOTING   HEADING   FLOWERING   MILKY   RIPENING  

PADDY  DIAGNOSTIC  ENGINE   PADDY  COMMUNITY  

PROJECT   PUBLICATION   MACHINE   TECHNOLOGY   INTELLECTUAL  PROPERTY  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

34  

Page 35: 2015 11-4-v14-ickm2015-keynote

HERBAL MEDICINE KNOWLEDGE REPOSITORY

HERBS  

TRADITIONAL  MEDICINE  PHARMACOLOGY  

PROCESS  PLANT  PART   ROOT   FLOWER  STEM  LEAF   SEEDS  

PLANT  ANATOMY  

HUMAN  ANATOMY  

FRUIT  2015  MIMOS  Berhad.  All  Rights  Reserved.  

35  

Page 36: 2015 11-4-v14-ickm2015-keynote

PADIPEDIA ONTOLOGY DESIGN

2015  MIMOS  Berhad.  All  Rights  Reserved.  

36  

Technology

isPreparedForRMK

hasTargetDateOfCompletion

xsd:date

xsd:integer

isCommercializedBy

isRelatedToCluster

hasCompletionDate

hasComDesc

hasComDate

hasPriority

hasRelatedProject

hasDevDesc

isRelatedToCommodity

dc:description

dc:title

rdf:type

PadiClassification

xsd:string

xsd:string

xsd:date

xsd:date

foaf:Project

Cluster

Commodity

xsd:string

xsd:string

xsd:integer

CorporateKPI

hasCorporateKPI

org:Organization

diagnose:hasMainSymptom

diagnose:hasPestDetail

agrovoc:hasScientificName

agrovoc:hasCommonName

diagnose:hasCharDetail

diagnose:isConfirmedBy

padi:hasManagement

diagnose:hasControlDetail

diagnose:hasStageDetail

xsd:string

agrovoc:PestLifeStage

rdf:type

EggLarvaPupaAdult

diagnose:destructAt

padi:occurAt

agri:isManagedBy

diagnose:hasPestCausal

xsd:string

xsd:string

agrovoc:Pest

diagnose:PestDetail

diagnose:MainSymptom

diagnose:RiceHealthDamage

agrovoc:GrowthStage

diagnose:affectsPlantPart

agrovoc:PlantPart

xsd:string

diagnose:hasSymptomChar

xsd:string

xsd:string

agrovoc:Management

agrovoc:Control

xsd:string diagnose:hasSymptom

xsd:string

diagnose:hasSymptomDetail

diagnose:hasSymptomImage

diagnose:hasLifeCycleDetail

diagnose:PestLifeCycle

diagnose:hasLifeStage

agrovoc:PestLifeStage

diagnose:SymptomDetail

diagnose:SymptomChar

diagnose:Symptom

diagnose:hasCharImage

xsd:string

diagnose:isContributed

Byfoaf:Person

diagnose:hasPestImage

Rice

growsIn

hasCroppingSystem

CroppingSystem

agri:hasCropProfile

agri:CropProfile

hasFamilyhasGenus

Family Genus

agri:hasProductionPractices

ProtectionTechnology

agri:hasSpecies

Species

agri:hasProductionPractices

ProductionTechnology

All properties have “agrovoc” namespace in frontunless different namespace is specified.

growsIn

Climate

influences

AgroclimaticZone

AgroEcoSystem

influences

padi:RiceEcosystem

hasComponent

padi:RainfedRiceCroppingSystem

padi:RainfedUplandpadi:RainfedLowland

padi:IrrigatedRiceCroppingSystem

padi:SingleCroppingSystempadi:DoubleCroppingSystem

SolarRadiationTemperatureRelativeHumidityRainfallDaylength

agri:ClimateComponent

Variety

padi:RiceVariety

hasVariety

Soil

skos:altlabel

xsd:string

agri:GeneralDescription

Origin

agri:hasOrigin

agri:hasImportance

agri:Importance

agri:NutritionalImportanceagri:StatisticalImportancepadi:SafetyAndHealthImportance

agri:hasGeneralDescription

padi:Graminae padi:

Oryza

padi:CultivatedSpecies

padi:OSativa

WildSpecies

padi:ORufipogonpadi:OOfficinalispadi:OGlaberimmapadi:ONivara

padi:RiceBasedCropRotationSystem

System

padi:AerobicRiceBasmatiVarietypadi:HerbicideTolerantVarietyHighYieldVarietypadi:JasmineVarietypadi:SpecialtyRicepadi:TraditionalRice

padi:HybridRice

padi:InbredRice

padi:RiverineAlluviumpadi:MarineAlluvium

RiceEcotype

padi:hasRiceEcotype

padi:Indicapadi:Javanicapadi:Japonica

MineralSoil

OrganicSoil

padi:ShallowOrganicSoilpadi:ModerateAndDeepOrganicSoilpadi:Agriculture

Importance

padi:SeedlingMediumpadi:SiliconeBasedFertilizerpadi:BioCharcoalpadi:SoilImprovementMaterialpadi:BioComposite

Page 37: 2015 11-4-v14-ickm2015-keynote

HERBAL MEDICINE ONTOLOGY DESIGN

2015  MIMOS  Berhad.  All  Rights  Reserved.  

37  

Herb

xsd:stringxsd:string

xsd:string

xsd:string

xsd:string

ChemicalConstituent

hasIPNINohasTROPICOSNo

hasSynonym

skos:prefLabel

hasDescription

xsd:string

hasChemicalConstituent

FamilyName

hasFamilyName

dcterms:Location

isOriginatedFrom

dcterms:Location

hasDistributedPlace

hasLink

hasReference

xsd:string

hasFloraOfChina

hasNaturalOccurance

dcterms:Location

hasTemperature

HerbTemperature

hasAnnualRainfall

propagatedBy

hasAltitude

HerbAltitude

PlantPart

SoilType

hasSoilType

Climate

hasClimate

hasRequiredFertilizer

hasDiseaseDetail

hasPestDetail

Fertilizer

DiseaseDetail

PestDetail

hasImageLink

hasVideoLink

xsd:string

xsd:string

hasSynonymLink

xsd:float

HerbAnnualRainfall

hasMinimumQuantity

hasMeasurementUnit

xsd:float

hasQuantity

hasMeasurementUnit

hasMinimumQuantity

hasMeasurementUnit

MeasurementUnit

hasMaximumQuantity

hasMaximumQuantity

PlantDisease

hasDisease

hasDiseaseControl

DiseaseControl

PestControl

PesthasPest

hasPestControl

xsd:string

HerbProduct

hasProduce

PlantPart

CommercialProduct

hasCommercialProduct

hasPlantPartUsed

causedBy

CausalAgent

xsd:floatxsd:float

xsd:float

MeasurementUnit

MeasurementUnit

hasDescription

xsd:string

hasDescription

xsd:string

herb:Herb

TraditionalMedicine

usedToTreat

HumanDisease

herb:hasPlantPartUsed

herb:PlantPart

hasMedicationPrescription

MedicationPrescription

herb:hasReference

HerbUsed

hasHerb

herb:hasMeasurementUnit

herb:Measurement

Unit

xsd:float

herb:hasQuantity

herb:hasQuantity

xsd:float

herb:hasMeasurementUnit

hasFrequency

hasTreatmentMethod

Activity

xsd:string

hasDurationDuration

Prescription

process:hasProcess

process:Process

has TraditionalMedicine

herb:hasLink

xsd:string

herb:hasQuantity

herb:hasMeasurementUnit

xsd:string

herb:Measurement

Unit

xsd:float

herb:Measurement

Unit

applyOn

BodyAreatradmed:

TraditionalMedicine

Process

expandTo

herb:hasDescription

xsd:string

followByhasProcess

hasInitialProcess

herb:Herb

ProducedChemical

Phamacology

tradmed:HumanDisease

hasProducedChemical

herb:hasMeasurementUnit

herb:Measurement

Unit

xsd:float

herb:ChemicalConstituent

herb:hasChemicalConstituent

herb:hasLink

hasPharmacology

hasReference

protectAgainst

hasPharmacologyToxicity

herb:hasQuantity

ChemicalToxicity

hasConcentrationUnit

useTestModel

TestModel

hasPharmacologyActivity

PharmacologyActivity

leadToToxicity PharmacologyToxicity

hasReference

xsd:string

hasChemicalExtract

useExtract

ExtractDetail

herb:hasQuantity

herb:hasMeasurementUnit

hasConcentrationUnit

ExtractTypehasExtractType

extractFromherb:PlantPart

hasTestModeldescription

herb:Measurement

Unit

xsd:string

xsd:float

xsd:string

xsd:string

herb:Measurement

Unit

herb:Measurement

Unit

Page 38: 2015 11-4-v14-ickm2015-keynote

Padipedia Demo

38  © 2015 MIMOS Berhad. All Rights Reserved.

Page 39: 2015 11-4-v14-ickm2015-keynote

SOCIAL MEDIA INTELLIGENCE

Page 40: 2015 11-4-v14-ickm2015-keynote

Challenges and Solution

40  

Internet  

Expensive,  Lengthy  and  subject  to  Error  

Manual  Searching  and  Monitoring    Automa*c  

Eliminate  the  manual  process,  Save  Tme  and  cost  

Aid  in  Decision  Making  

Centralized  Analy*cs    

Dashboards    

© 2015 MIMOS Berhad. All Rights Reserved.

Page 41: 2015 11-4-v14-ickm2015-keynote

Social Media Analytics

Internet  

3.  Visual  reports  on  Insights  presented  to  decision  maker  for  further  ac8on  

1. Harvest  content  from  the  internet  (world  wide  web,  social  web,  content  web)  

Automated  Data  HarvesTng  2.  Social  Media  Analy8cs  to  

generate  insights  about  the  topic  of  interest  

Social  Media  Analysis  

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

Domain  Ontology  41

Page 42: 2015 11-4-v14-ickm2015-keynote

Mi-Intelligence Process Overview

42  

Expansion  

HarvesTng  

Processing  

SubjecTve  AnalyTcs  

Visual  AnalyTcs  

Domain    Ontology  

Insights  

GATHER    DATA  

SEMANTIFY    DATA  

DISCOVER    KNOWLEDGE  

IDENTIFY    PATTERNS   Domain    

Ontology  

ESTABLISH  SEARCHSPACE  

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

42

Page 43: 2015 11-4-v14-ickm2015-keynote

Automated Data Harvesting

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

1. Harvest  content  from  the  internet    (world  wide  web,  social  web,  content  web)  

Add  new  topics  to  analyse  

Summary  of  data  harvested  and  processed  

Search  topics  

43

Page 44: 2015 11-4-v14-ickm2015-keynote

Content Insights

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

Where  are  the  posts  coming  from?  

Drill  down  to  discover  detailed  informaTon  

about  posts   Who  are  the  users?  

Where  are  the  posts  coming  from?  

44

Page 45: 2015 11-4-v14-ickm2015-keynote

Social Media Analysis

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

Sen*ments  

Domain  Ontology  

Emo*ons  

Anxiety  machine  understandable  

domain  knowledge  

45

Page 46: 2015 11-4-v14-ickm2015-keynote

Social Media Analysis

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

2.  Social  Media  Analy8cs  to  generate  insights  about  the  topic  of  interest  

Sub  topics  related  to  the  main  topic  

SenTments  from  social  media  

SenTments  from  different  regions  

SenTments  from  different  regions  

46

Page 47: 2015 11-4-v14-ickm2015-keynote

Social Media Analysis

©  2015  MIMOS  Berhad.  All  Rights  Reserved.  

2.  Social  Media  Analy8cs  to  generate  insights  about  the  topic  of  interest  

Drill  down  to  discover  detailed  informaTon  

about  posts  

SenTments  from  different  regions  

SenTments  from  different  regions  

Narrow  down  to  the  individuals  posts  of  

interest  

47

Page 48: 2015 11-4-v14-ickm2015-keynote

Mi-Intelligence Demo

©  2014  MIMOS  Berhad.  All  Rights  Reserved.  

48  

Page 49: 2015 11-4-v14-ickm2015-keynote

SOCIAL NETWORK INTELLIGENCE

Page 50: 2015 11-4-v14-ickm2015-keynote

2015  MIMOS  Berhad.  All  Rights  Reserved.  

Harvest,  Extract,  Harmonize,  and  Transform  Data  

Informa8on  related  to  Suspects  

Drugs  Cases  

JKDM  

WCO  

Rule  Mining  (Machine  Learning)   SME  

(Risk  Officer)  

JKDM  HQ  

Suspect  KB  

Rules  

Social  Network  IdenTficaTon  (Mi-­‐Target)  

750  Suspects  30  Million  Facts  

Suspect Data Preparations

50

Page 51: 2015 11-4-v14-ickm2015-keynote

Suspect Network Analysis

Harvest,  Extract,  Harmonize,  and  Transform  Data  

Informa8on  related  to  Suspects  

Suspect  Network  Analysis  Engine  (Mi-­‐Visuali*cs)  

Poten*al  High  Risk  Individuals  •  Most  InfluenTal    •  Top  Connectors  

Intelligence  Officers  

         

Suspect  Watchlist  

Drugs  Cases  

JKDM  

WCO  

Suspect  KB  

Intelligence  Officers  

Other  Informa*on  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

51

Page 52: 2015 11-4-v14-ickm2015-keynote

2015  MIMOS  Berhad.  All  Rights  Reserved.   52  

Rule  Mining  (Machine  Learning)  

SME  (Risk  Officer)  

Passenger Risk Profiling

Adap*ve  Personal  

Demographic  Analysis  

(Mi-­‐Target)  

Observa*on:  •  Travelling  Purpose  •  DuraTon  of  Stay  •  Profession  •  Physical  Appearance  

Flight  Informa*on:  

• Flight  Code  • Airline  Operator  • DeparTng  Airport  • Arrival  Airport  

Personal  Informa*on:  

• Name  • Passport  Number  

• NaTonality  • Age  

         

Suspect  Watchlist  

JKDM  HQ  

Airports  

Suspect  KB  

Rules  

Page 53: 2015 11-4-v14-ickm2015-keynote

Profiling Rules using Ego-Group

Intelligence  Officers  

         

Suspect  Watchlist  

Hector  B.L  

Hector  Beltran  Leyva  

Adap*ve  Personal  

Demographic  Analysis    

(MI-­‐Target)  

Suspect  KB  

Rules  

Suspect  Network  Analysis  Engine  (Mi-­‐Visuali*cs)  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

53

Page 54: 2015 11-4-v14-ickm2015-keynote

Mi-Visualitics Demo

54  © 2015 MIMOS Berhad. All Rights Reserved.

Page 55: 2015 11-4-v14-ickm2015-keynote

CONCLUSION

Page 56: 2015 11-4-v14-ickm2015-keynote

Where was Semantic Technology applied in Knowledge Management ?

•  Knowledge Harvesting (LOD) •  Social Media Harvesting & Filtering •  Knowledge Asset Organization &

Management •  Data Harmonization •  Text Understanding •  Natural Language Question Answering •  Semantic Search •  Social Media Intelligence •  Social Network Intelligence •  Knowledge Management Portal

56  © 2015 MIMOS Berhad. All Rights Reserved.

Page 57: 2015 11-4-v14-ickm2015-keynote

ありがとうございます  

Page 58: 2015 11-4-v14-ickm2015-keynote

Knowledge Audit

2015  MIMOS  Berhad.  All  Rights  Reserved.  

58  

Enterprise  Data  Linked  Open  Data  

Sensor  Web  

ACTIONABLE  KNOWLEDGE  

ASSETS  

Structured,  Semi-­‐Structured  &  Unstructured  

Unstructured   Structured  &  Semi-­‐Structured  

Page 59: 2015 11-4-v14-ickm2015-keynote

Ontology

Technical  WriTng  

Term  Databanks  

Machine  TranslaTon  

Human  TranslaTon  

InformaTonRetrieval  

Knowledge  Engineering  

Consumer  InformaTon  

R&D  

StandardisaTon  

Nomenclature  

Terminology

AGROVOC

MYGMO  

PADIPEDIA  

AGRIS  

HERBAL  MEDICINE  

2015  MIMOS  Berhad.  All  Rights  Reserved.  

59  

SNOMED-CT STW