4. Scarecrow : I haven't got a brain... only straw.
Dorothy : How can you talk if you haven't got a brain?
Scarecrow : I don't know... But some people without brains
do an awful lot of talking... don't they?
Dorothy : Yes, I guess you're right
Scarecrow : The sum of the square roots of any two sides
of an isosceles triangle is equal to the square
root of the remaining side. Oh joy! Rapture! I
got a brain! How can I ever thank you enough?
Wizard of Oz: You can't.
Communicating Knowledge 3
5. Tony’s Brain and Knowledge
Neurons
~100B #
~2x # of
Web Pages
Synapse
~100T #
~2# of
Web Links
Communicating Knowledge 4
15. Data, Information and Knowledge
DATA
• symbol, a statement
• facts of the world
INFORMATION
• collection of data, data in context
• answer about who, what, where, when
KNOWLEDGE
• contextual organization of information
• map of the world inside our brains
By Gene Bellinger, Durval Castro and Anthony Mills • answer about how and why
Communicating Knowledge 14
16. Knowledge Representations
Human
Natural Language Human language written in letters: “The Earth orbits the sun in an ellipse”
Visual expression of knowledge in picture, structure diagram, flow chart,
Visual Language and blueprint etc
Knowledge expressed in keywords, symbols and images related with
Tagging objects
Symbolic Language Knowledge expressed in mathematical symbols : x2/a2 + y2/b2 = 1
Decision Tree Tree-shaped graph structure for complex decision making
Combined expression in condition with various rules of human
Rules Language knowledge
Knowledge expression system composed of objects and relations in a
Database System table format
Knowledge expression of logical symbols and arithmetic operations:
Logical Language
Machine
Woman ≡ Person ∩ Female
Knowledge expression of values or pointers for other frames saved in
Frame Language slots
Knowledge expression of semantic relation between concepts in a graph
Semantic Network structure
Allows knowledge expression, machine learning technology combination
Statistical Knowledge based on probability and statistics
Communicating Knowledge 15
17. Knowledge Representations
Natural Language
“Employees working for a company are humans; the company and the employees are legal
entities. The company is able to make a reservation for an employee’s trip. The trip is
available by plane or train that travels in cities within Korea or the U.S.. The companies
and destinations for business trip are located in the cities. Saltlux reserved OZ510 with a
round trip of Seoul and New York for Hong, Kildong.”
Rule Language
(Rule) If someone is flying, he must be on trip.
(Rule) If someone’s trip is reserved in a company, he is an employee of the company.
(+ Rule) For short trip in the same country, an employee should take a train.
(Deduction) Hong kil-dong whose flight is in reservation is an employee of Saltlux.
(Deduction) OZ510 is a flight for the U.S. and Korea.
Communicating Knowledge 16
18. Knowledge Representations
Legal Entity Legal Entity
Name Location
Name (*)
Legal Entity ID ID (*)
Legal Entity
kindOf
DISJOINT
Person Company startFrom Person Company
Gender Industry Gender⊆{M,F} Industry
Person Company Company Age
Person books Trip
Address Age > 25
City
Addr⊂Seoul
endsIn
subclssOf
subclssOf
subclssOf
Ontology
kindOf
instanceOf
instanceOf
instanceOf
instanceOf
Employee Employee
Korean American
Airplane
Position Train Pos ≠ Exec.
Employee
Employee City City
instanceOf
instanceOf
#4831 #4831
instanceOf
instanceOf
instanceOf
Saltlux Saltlux
instnaceOf
instanceOf
Saltlux
Saltlux C98765 C98765
#3502 Software #3502 Software
Seoul Seoul
Kildong Seoul
Kildong
Kildong participatesIn
P12345 P12345
Kildong OZ510
Male Male
37 37
New york
Manager Manager
(a) Semantic Network
Semantic Frame (Slots)
(b) (a) +
Network (c) (b) + Logical Restrictions
Communicating Knowledge 17
19. Five View Points for Semantic Technology
• URI/RDF based “Web of Data”
• Semantic annotations (RDFa)
• Ontology and Logics • Reasoning, Agent system
• OWL and RIF • Personalized services
• Linked semantic data • Semantic Search and Mining
• Data interoperability • Recommendation and Discovery
Communicating Knowledge 18
22. Hybrid Reasoning : (1) Mixed Method
Logical Reasoning Methods
Ontology and Rules
• Deductive reasoning
Premise 1: All humans are mortal.
Premise 2: Socrates is a human.
Conclusion: Socrates is mortal.
+
• Inductive reasoning
Premise: The sun has risen in the east every morning up until now.
Conclusion: The sun will also rise in the east tomorrow.
• Abductive reasoning Machine Learning
• Analogical reasoning
Communicating Knowledge 21
23. Hybrid Reasoning : (2) Mixed Formalism
+
The relationships among different formalisms Semantic Web Architecture
(Benjamin Grosof)
Communicating Knowledge 22
27. GEO Data, GEO Information and GEO Knowledge
WHAT IS
GEO-
GEO-KNOWLEDGE ?
www.ci.ferndale.wa.us/ GIS/ GIS.php
Communicating Knowledge 26
28. An Evolution of Geo Ontology
Geo Tagging
GPS based POI processing
Connecting location coordinate with the relevant
information
Geo Features
Applying classification system by domain
Referring to major geographic classification system
such as GeoNames
Geo Ontology
Building/Applying ontology-based spatial information
Expressing Point/Line/Shape information
Geo Ontology + Rules
Utilizing Geo Ontology and rule-based inference
Applying deduction rules for intelligent spatial
information processing
Communicating Knowledge 27
33. POI and Geo-data modeling by Saltlux
• 28# main category and 600 sub categories for POI classification
• Using SKOS for semantic classification
• Including named places and events
• All data set has its name space, http://www.saltlux.com/geospatial
• Coordination
- http://www.w3.org/2003/01/geo/wgs84_pos#lat
- http://www.w3.org/2003/01/geo/wgs84_pos#long
SKOS based Taxonomy GEO Data Model
http://www.w3.org/2004/02/skos/core#ConceptScheme http://www.w3.org/2004/02/skos/core#Concept
http://www.w3.org/2004/02/skos/core#broaderTransitive
http://www.w3.org/2004/02/skos/core#narrowerTransitive
http://www.w3.org/2004/02/skos/core#broader
http://www.w3.org/2004/02/skos/core#narrower
http://www.w3.org/2004/02/skos/core#hasTopConcept
http://www.w3.org/2004/02/skos/core#inScheme
http://www.saltlux.com/geospatial#Class http://www.saltlux.com/geospatial#Code
Communicating Knowledge 32
34. Taxonomy and Property modeling for POI
NO Main Classes Sub Classes name good for
1 Arts & Entertainment Arcades
Arts & Entertainment Adult Entertainment & Nightlife
Adult Arcades, Casinos Archery
Sports & Recreation alternate name products and services
Art Galleries Adult Massage Badminton
2 Adult Entertainment & Nightlife Botanical Gardens, Arboretum Adult Novelties & Product Shop Baseball description specialities
Cinema Bars & Pubs Basketball
Music Concert Hall Business Clubs Billiards
3 Sports & Recreation Open air theatres, Festival Places Audult Comedy Clubs Boating street-adress brands
Theatres Dance Clubs Bowling
4 Media & Broadcasting Museums
Astrologers & Psychics
Jazz & Blues Clubs
Audalt Karaoke
Boxing
Cricket postal-code smoking
Social & Interests Clubs Hourse Racing Curling
5 Religious Organizations Talent Agencies & Entertainers Boat Racing Cycling
categories take-out
Party Rentals Bycycle Racing Dance
Ticket Office Car Racing Equestrian
6 Transportation Aquariums Billiard Halls Fencing
url transit
7 Automotive
8 Education & Learning
Video/DVD, Game Rental
Comedy Clubs
Circus
Amusement Parks
Zoos
Other Adult Entertainments Fishing
Fitness Clubs
American Football
Golf
Gun/Rifle Ranges
… email wireless
Cartoon Rental Gymnastics tel reservations
9 Event Planning & Services Internet Café
Karaoke
Hockey
Horse Racing & Equestrian
lottery shop Hunting latitude best nights
10 Manufacturing & Industry General Exhibition, show Lacrosse
Folk Village Taekwondo
11 Financial & Legal Services Other Arts Judo longtitude alcohol
Other Entertainment Kendo
Martial Arts
12 Health and Medical Motorsports & Racing device_lat reviews
Paintball
Parachuting
13 Beauty and Spas Racquetball device_long photo
Rafting/Kayaking
14 Other Professional Services Ringette
Rugby
hours Year Established
Running
15 Travel & Tours Scuba
fax seating
Skateboarding, inline skating
16 Home & Local Services Ice Skating
Skiing
Skydiving
payment options outdoor seating
17 Shopping Soccer
Softball parking chef
18 Government & Public Services Squash
Surfing
Swimming ambiance self service
19 Food & Drink Tennis
Track & Field
Volleyball amentities languages spoken
20 Restaurants Wrestling
Aerobic
21 Other Artifacts Yoga attire music
Bungee Jump
22 Other Natural Objects
Camping
Arenas & Stadiums
price range associations
Playground
23 Utility & Infrastructure Other Sports
delivery part_of
Taxonomy Modeling Property Modeling
Communicating Knowledge 33
35. Semantic queries for geo-data
Find good restaurants for dating that serves steaks and
free parking near by GAGA gallery in Insa-dong.
PREFIX ns: <http://www.saltlux.com/geospatial#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX dc: <http://purl.org/dc/elements/1.1/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
PREFIX wgs: <http://www.w3.org/2003/01/geo/wgs84_pos#>
PREFIX f: <http://www.saltlux.com/geo/functions#>
SELECT *
WHERE {
?res rdf:type ns:NamedPlace;
ns:name ?name;
wgs:lat ?lat1;
wgs:long ?long1;
ns:address ?addr1.
OPTIONAL{?res ns:street-address ?straddr1.}
?rel rdf:type ns:NamedPlace;
ns:name ?relname;
wgs:lat ?lat2;
wgs:long ?long2;
ns:address ?addr.
OPTIONAL{?rel ns:street-address ?straddr.}
?rel ns:category ?cate;
ns:ambiance ns:ambiance_129;
ns:parking ns:parking_9.
?cate rdfs:label ?catename.
FILTER (f:distance(?lat1, ?long1, ?lat2, ?long2) <= 300 && ?cate =
<http://www.saltlux.com/geospatial#code_655> && ?name = '가가갤러리')
} ORDER BY ?relname
Communicating Knowledge 34
37. MOBILE APIs
• Supporting Mobile APIs by using SPARQL Endpoint
• Working on Android (and iPhone)
function description
findPOIbyAll(java.lang.String name, java.lang.String id, int dist, 4 argument(name, id, distance, format)
java.lang.String format)
format: xml, json, id: Category ID
findPOIbyAllCoordinate(java.lang.String id, int dist, double lat, 5 argument(id, distance, lat, long, format)
double lon, java.lang.String format) id: Category ID, format: xml, json
findPOIbyCoordinate(java.lang.String id, int dist, double lat, 4 argument(id, distance, lat, long)
double lon) id: Category ID, format: xml
findPOIbyDist(java.lang.String name, java.lang.String id, int dist) 3 argument(name, id, distance)
name: name of PIO, id: Category ID
findPOIbyName(int dist, java.lang.String name) 3 argument(distance, name)
format: xml
findPOIbyFormat(int dist, java.lang.String name, 3 argument(distance, name, format)
java.lang.String format) format: xml, json
findPOIbyID(java.lang.String id) 1 argument(id)
id: 특정 상점이 갖는 URI, return: meta data of POI, xml
geoQuery(java.lang.String query, java.lang.String format) 2 argument(query, format)
format: xml, json
query for getting MAP data
query(java.lang.String query, java.lang.String format) 2 argument(query, format)
format: xml, json
SPARQL query
Communicating Knowledge 36
38. MOBILE APIs
protected void onStart() {
super.onStart();
// Service binding
Intent i = new Intent(this, SparqlEndpoint.class);
boolean ret = bindService(i, mConnection, Context.BIND_AUTO_CREATE);
private ServiceConnection mConnection = new ServiceConnection() {
// Service binding call
public void onServiceConnected(ComponentName name, IBinder service) {
// converting from service into ISparqlEndpoint interface
mService = ISparqlEndpoint.Stub.asInterface(service);
mService.query( … );
}
// Closing service
public void onServiceDisconnected(ComponentName name) {
mService = null;
}
};
…
}
Communicating Knowledge 37
39. Use-case : Urban Computing in LarKC project
Wikipedia [DB] [Triple] [DB]
DBPedia LinkedGeoData OpenStreetMap
(LGD)
http://www.larkc.eu Mapping with Triplify Making XML data to Linked Data
Owl:SameAs,
Jaro distance metric
Name comparing
Schema : 24Class
Element : 320mega Nodes,
25mega Ways
Communicating Knowledge 38
40. Use-case : Urban Computing in LarKC project
Inconsistency check
• Some POIs suddenly disappeared at a point and
reappeared at another point which make arrow
directions are inconsistent.
Unsuitability check
• some of road signs are not properly designed
and installed as specified by the regulation.
• improperly located road signs such as those not
installed within the specified range from junction
Inconsistent road sign Unsuitable road sign points are not suitable one.
Continuous planning
• Some of POI are destroys and moves or appears
from time to time. One of major POI
movement(for example, city hall) may bring many
road sign modifications.
Go straight? Or turn right? New road sign covers another
road sign
Communicating Knowledge 39
41. Use-case : Urban Computing in LarKC project
[Node Element]
[Road Sign & POI Insertion]
R K E
R
Node RS KPOI WPOI
[Link Element] K
E
startNode endNode sameAs
[Junction & POI Finding]
[Way Element] R2
R1
nextLink
sameAs
J1
J2
K
link N E
link M
[Road Element] K
way N
K
E
way M
searching range
Communicating Knowledge 40
42. Use-case : Urban Computing in LarKC project
Data Set Comments
Linked Geo Data 1 billion triples in WGS84 coordinate
(LGD) Loading LGD full and extracted part of Korea
Open Street Map Extracting all way information in WGS84 coordinate
(OSM) Selecting and importing 2 million triples for Seoul
Point Of Interest
1 million POIs related with road signs
Data in Korea
Around 4 million triples.
(KPOI)
Diverse data of Seoul road sign in database
Seoul road sign 9515 instance of direction road signs in Seoul
data (RSD) Converting TM coordinate into WGS84 coordinate
Converting RDB into RDF (0.5 million triples)
Korean road sign Around 30 Regulations of road signs
regulations (RSR) Changing to SparQL for validation check
Mediate Ontology Ontology linking between OSM, KPOI, RSD or other data
(MO) Expressivity : subClassOf, subPropertyOf, sameAs, inverseOf
Total : about 1.1 billion triples
Communicating Knowledge 41
45. Use-case : Urban Computing in LarKC project
Finding the target POI around 500m from a node of road
PREFIX rsm: <http://www.saltlux.com/rsm#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
PREFIX sgf: <http://www.saltlux.com/geo/functions#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX geo: <http://www.w3.org/2003/01/geo/wgs84_pos#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
select distinct ?targetPOIID
where {
rsm:osmn_436718764 geo:lat ?endNodeLat .
rsm:osmn_436718764 geo:long ?endNodeLong .
rsm:kpoi_12720 geo:lat ?targetNodeLat .
rsm:kpoi_12720 geo:long ?targetNodeLong .
rsm:kpoi_12720 rsm:id ?targetPOIID .
filter ( sgf:distance(?endNodeLat, ?endNodeLong, ?targetNodeLat, ?targetNodeLong) <= 500 )
}
Communicating Knowledge 44
46. Use-case : Urban Computing in LarKC project
Data: Traffic Flow and Speed Prediction: Data from Milano
Milano City Sensor Map
Traffic data from Milano (Italy)
Data ranging from Mar. 07 to July 09
(849 days)
5 min. sampling rate for flow & speed
Traffic flow & speed from
209 sensors that are able to
classify vehicles, and
757 non classifying sensors
Weather data provided from
http://www.ilmeteo.it
1 hour sampling rate for weather data
Sensors – Crossroads – Street Categories (multi-colored)
Communicating Knowledge 45
47. Use-case : Urban Computing in LarKC project
Problem Description: Traffic Flow and Speed Prediction
Traffic Flow [12:00; 12:05]
(preprocessed)
Traffic Flow (# vehicles)
Mar. 07 July 09
Traffic Speed [12:00; 12:05] Predict the traffic flow and
(preprocessed)
speed for the next 24h based
Traffic Speed (average)
on a 5 min. time grid
Traffic flow and speed forecasts
are made on the sensor level
for the whole traffic network
Forecasts: inputs for optimal
Mar. 07 July 09
routing algorithms
Communicating Knowledge 46
48. Context Awareness and Geo-Semantics
Source : Flickr.com, David Crow
Communicating Knowledge 47
49. Context Awareness and Geo-Semantics
• physical contexts
: location, time
• environmental contexts
: weather, light, sound levels
• informational contexts
: stock quotes, sports scores
• personal contexts
: health, mood, schedule, activity
• social contexts
: group activity, social relationship
• application contexts
: e-mail, websites visited
• system contexts
: network traffic, status of printers
Communicating Knowledge 48
50. Ontology for Context Awareness by Saltlux
Time
subClassOf subClassOf subClassOf
EventRelative
PersonRelative TimeInterval TimeRelation TimeWordMean subClassOf subClassOf
Profile Health
date before timeWord Event EventTerms
privacyLevel privacyLevel
disease time at before terms timeInterval
name
age bloodType after at actionTask person
LocationSensor dateInterval
birth heartBe device
timeInterval during after
job Person at
height Person deivceName
weight from serviceInterval
personID Device operation
Contact bodyTemperature to
locationInfo order
privacyLevel hasProfile etcHealthInfo
1
Session Task SubTask space
cellularPhoneNum contact etcHealthInfo
2
email person taskName subClassOf subClassOf subClassOf
health
hasNick Nick hasTask hasSubTask SearchSubTask DeviceSubTask OperationSubTask
Interest
privacyLevel belongTo nickName timeInterval searchWord device targetProperty
field preference calledBy Space space timeInterval deviceName propertyValue Space
detailedField interest
Group taskPriority delayTime
deviceAccessLevel hasOperation subClassOf subClassOf
groupName
privacyLevel timeInstance 이미지를 표시할 수 없습니다 . 컴퓨터 메모리가 부족하여 이미지를 열 수 없거나 이미지가 손상되었습
니다 . 컴퓨터를 다시 시작한 후 파일을 다시 여십시오 . 여전히 빨간색 x가 나타나면 이미지를 삭제한 다
hasMember HomeSpace
음 다시 삽입해야 합니다 .
OfficeSpace
userPriority Policy
subClassOf subClassOf delayTime
hasSchedule spaceName
registerService Level Priority
Person owner
subClassOf subClassOf subClassOf subClassOf
LocationSensor locationSensor
DeviceAccessLevel PrivacyLevel UserPriority TaskPriority
etcSensor etcSensor
level level level level
EntityRelative Device device
Actuator UserCommandRelative
Sensor
OperationLevel Display subClassOf
subClassOf subClassOf Device
operationLevel mode
deviceID UserCommand
LocationSensor EtcSensor Temperature brightness subClassOf
IPaddress controlColor personID Schedule
sensorID sensorID degree
deviceName contras macID Person who
coordinate coordinate humidity TaskType
actuator menuDisplayTime IPAdd title
sensingDataType sensingDataType Light subClassOf subClassOf subClassOf
deviceDefault Sound requestTime where
sensingTime sensingTime bright
color volume O
hasServiceTime DeviceTask RequestInformation BatchTask withWhom
status status owner
soundType from
hasSensor Communication OhasTaskType deviceID queryLiteral taskData
modeEQ to
alternativeDevice deviceName nodeType taskName
Power briefingContents
deviceAccessLevel Player O
hasOperation catagory
Exclusive powerStatus TimeRelation
coordinate track moviePlace
device Broadcast from Operation
playerStatus
remainBattery station to targetProperty
isWork OpenClose channel at propertyValue
ocStatus
Communicating Knowledge 49
51. Concept of Context Driven Mobile Services
CONTEXT MANAGER CONTEXT
SENSOR / NETWORK
QoC Inferred
Context Model
Context Rules
Context
CONTEXT OWNER Filter
Dynamic
Collector Context
User Device
CONTEXT-AWARE SERVICE
Service Service Service
Discovery Personalization Adaptation
Smart Mobile Service
Communicating Knowledge 50
57. Use-case : Trajectory Awareness
Hybrid Reasoning : DL Reasoning + Induction(Machine Learning)
Home Detection
SPARQL Query and Selection Clustering Calculation
DL Reasoning
Workplace Detection
GPS Log Selection Clustering Calculation
Data
Bus Stop Detection
Segmentation Cleansing Matching
Communicating Knowledge 56
58. Use-case : Trajectory Awareness
Hybrid Reasoning : DL Reasoning + Induction(Machine Learning)
Main Line Extracting Commute Trajectory Learning
Sampled
MobileLog Line Line Line Trajectory Trajectory Trajectory
Data Generating Thickening Thinning Graph Tracing Clustering
E1 E2 E6 E7
Thu Nov 18 09:52:35 13 min
Fri Nov 19 09:22:32 10 min
E6
E6
Sat Nov 20 09:44;30 13min
E7
V4
V4
E7
Thu Nov 23 08:18;39 6min
V5
V5 E4
E2
E2
E4
V6
V6 Wed Nov 24 12:35:52 15min
Thu Nov 25 09:31:23 14 min
E5
E5 V3
V3
Fri Nov 26 09:38:21 10 min
V2
V2 V1
E3
E3
V1
E1 E3 E5 E7
Wed Nov 17 08:06:19 9min
E1
E1 Mon Nov 22 08:06:34 10min
Sat Nov 27 09:28:27 14min
Communicating Knowledge 57
59. Use-case : ITS of u-City
1. No Accident and Disaster
2. Finding Accident and Disaster : Recommending Detour Path
3. Finding Accident and Disaster but it could be recovered soon
Communicating Knowledge 58
61. Use-Case : Water and Gas Pipeline Management
Context-Aware Disaster Management POC by Saltlux
System Overview
Context Aware System
• 4 System blocks
Service Manager Situation Reasoning
• OS : Windows Server
Service
Query Engine Triple Store • Platform : J2EE based POJO
Interface
Service (Pure Object Java Object)
Reasoner Rule Store
Handler • Supporting Web service
Context Acquisition Context Manager
Context Filter Instance Population
Context Collector Instance Manager
Disaster Control
Center Web Center
Context Service
Sensor GIS
Data Layer Logs Dash
Data Data Aware
Board
System
Communicating Knowledge 60
62. Use-Case : Water and Gas Pipeline Management
Context-Aware Disaster Management POC by Saltlux
Sensor Monitoring Leakage Detection Discover Leakage Area
Infer Leakage Pipe Link Automatic Alert Recom. Detour Path
Communicating Knowledge 61
64. Characteristics of Social Network and Networking
1. Structural features
• Small world : six degree of separation
• Unfair world : governed by power law
• Strength of weak relationship
2. Synchronization and Amplification
3. Empowerment in Network
Communicating Knowledge 63
65. Structure : Small World
• Six degrees of separation
- Experiment : Six degrees of
Kevin Bacon
• Does internet and SNS
make it shorter?
- Yes or No ?
- Average degrees in FB : 5.73
- Maximum degree in FB : 12
- Average degrees in TW : 4.67
- Average degrees in WP : 4.5
Communicating Knowledge 64
66. Structure : Unfair World
• Scale-Free Network
- Governed by power law
- Like Pareto and long tail principle
• Evolving to HUB network
- Portal(google), SNS(FB, TW)
- Airport, logistics(Fedex)
• Unfair world, reality
- men-women network
- The rich get richer and the poor
get poorer
Communicating Knowledge 65
68. Strength of Weak Relationship
• Strong and Closed Network
- Mafia organization, trust network
- no secret, no new information
• Weak and Open Network
VS. - a broker among heterogeneous nets
- controller of network and info. flow
• Strength of Weakness
- multiple and cross discipline become
more important
- getting new job and ideas, building
new business and innovation
Communicating Knowledge 67
71. Empowerment and democracy in Social network
Power Law VS. Power Dispersion
Communicating Knowledge 70
72. Semantic Social Network Analysis
• Social Networks : networks based on the relation between people
• Semantic Social Network : RDF representations of social network and data
Abstraction stack for semantic SNA
Foaf:knows
Foaf:interest
[Paolillo and Wright 2006]
Rich graph representations reduced to simple [Semantic Social Network Analysis,
untyped graphs in order to apply SNA http://journal.webscience.org/141/2/websci09_submission_43.pdf]
71
Communicating Knowledge 71
73. Semantic Social Network Analysis
Semantic Network Social Network
Text Mining SNA
(Induction) (Deduction)
Semantic Social Network Analysis
Communicating Knowledge 72
74. Semantic Social Network Analysis
People
Task Org.
Semantic
Network
Place Service
Event
Dijkstra’s algorithm
Family, Colleagues Information Connecting
vk : weighting of relation
n :Community
, number of relations gjk : # ofHub, Broker
shortest path between j and k Experts
g : total number of entity gjk(i) : # of path having i between j and k O( | E | + | V | log | V | )
73
Communicating Knowledge 73
75. SAMZZIE : Social Semantic Platform by Saltlux
Web User Interface
Knowledge Network Services
User User & Admin Knowledge
Social Knowledge Knowledge Contents K-Dic DAC
Type Topic Rank Knowledge Discovery
Network Trend Circulation Search Manager Schedule
Pattern Env. Query
Integrated API
Knowledge Network Analysis Knowledge Discovery Control
KNA KDC
Knowledge User Type
Social Network Knowledge DA DAR KNA
Circulation Pattern
Analysis Trend Analysis Scheduler Scheduler Scheduler
Analysis Analysis
Data Aggregation Data Analysis and Reasoning
DA DAR TMS TRE
Topic
Document NE & Data Knowledge Query Feature
Email Aggr. Web Aggr. Ranker
Aggr. Annotation Abstraction Population Engine Extraction
Engine
Meta Base Knowledge Base
Email Topic Topic Email U&A DA K-Dic KDQ Authority Schedule SN KT KC UTP
Contents Trend Rank Abst. Env. Policy
Communicating Knowledge 74
78. Use-case : Intelligent Telco Platform
Intra Portal Solution Intelligent Application Service (AS)
Personalized Intelligent Shopping Social
FIMM MagicN Local News Etc.,
Search Traveling Guide Recommendation Network
Internet Portal,
Legacy System CDE Enabler
BcN, All-IP N/W
MagicN
VME
ISMS ICDS empas
IPAS
ICE
AAA SICS O&M DAUM
PS
SAS
NAVER
JUICE
Simulation
Server Internet
HSS TV
S(L)MSC
AuC P/I/S-CSCF
BGCF BcN
S(L)MSC
IMS-ALG All-IP
IMS Infra TrGW Network
PCRF
• ISMS: Intelligent Subscriber information Mgnt. Server
Node-B • ICDS: Intelligent Content Delivery Server
RNC SGSN GGSN
(WCDMA) • SICS: Subscriber Information Collection Server
W-CDMA • O&M: Operation & Management Server
Communicating Knowledge 77
79. Use-case : Mobile Social Network Analysis
Major Activity Area
Major Residential Area
attend
Profile
attend
• Name: Jerry
Profile
Obama
• Age: 12
• Name: Elizabeth attend
lives in • Sex: Woman
Cox
lives in • Age: 12
Pay for • Sex: Woman attend attend
Call Profile
• Name: Jane Bush
• Age: 12
Call • Sex: Woman
Call
Call
Profile
Profile
Call
• Name: Edward
• Name: Nancy
Adams
Obama
Call
• Age: 11
• Age: 42
• Sex: Woman
• Sex: Woman
lives in
Profile
Profile
• Name: Jessica
• Name: Tom Obama
Bailey
Family • Age: 16
Friends • Age: 13
• Sex: Man
• Sex: Woman
Communicating Knowledge 78
80. Use-case : Personal Profile Analysis
User Profiling from Network
Bimodal
Normal
μ1 = 38 Φ=
σ1 = 4.2
w = 0.83 각 나이대별 (SMS/VOICE), (성별),
μ2 = 13
σ2 = 2.4 (나이대) 120 클래스 패턴을 이용한 PI
결과
Communicating Knowledge 79
81. Use-case : Personal Profile Analysis - Semantics
Ontology Population
Legacy Data
Ontology Mapping from Legacy Data
80
Communicating Knowledge 80
91. Hybrid Reasoning and Queries
tweets about a given kind of POI of people similar to me that tweeted nearby in the last x minutes;
PREFIX f: <java:ext.>
SELECT ?poi1 ?poi2 ?user (f:similarWithProbability(data:Alice, ?user) AS ?p)
WHERE {
?user bot:posts ?t1 .
?t1 bot:talksAboutPositively ?poi1 .
?poi1 a bot:NamedPlace ;
geo:lat ?lat1 ;
geo:long ?long1 ;
skos:subject ?category .
data:Alice geo:lat ?givenLat ;
geo:long ?givenLong ;
bot:posts ?t2 .
?t2 bot:talksAboutPositively ?poi2 .
?poi2 a bot:NamedPlace ;
geo:lat ?lat2 ;
geo:long ?long2 ;
skos:subject ?category .
FILTER(?t1!=?t2)
FILTER(f:similarWithProbability(data:Alice, ?user)>0.5)
FILTER((?lat1-?givenLat)<"0.1"^^xsd:float &&
(?lat1-?givenLat)>"-0.1"^^xsd:float &&
(?long1-?givenLong)<"0.1"^^xsd:float &&
(?long1-?givenLong)>"-0.1"^^xsd:float )
}
ORDER BY DESC(?p)
LIMIT 10
Communicating Knowledge 90
92. BOTTARI mobile App by Saltlux
AR based Location Search Reputation Analysis
Intro screen Social Recommendation Expert Search and
and Dynamic Social search Real-time Q&A
Communicating Knowledge 91
96. Future Use-case : Disaster Management
Mobile phone Sensors Social Media & Networks Satellite/CCD images Sensor Networks
Stream Data Hybrid Stream Reasoning
Massaging Service
Collection
Decision Supporting
Geo-Spatial
Data Collection
Citizen
Decision Dashboard
Disaster Disaster Disaster Supporting And
Knowledge Process Policy System Controller
Government ,
Disaster Center
GIS and Geo-Data
Intelligent Disaster Management System
Communicating Knowledge 95
97. Technical Tips
And the Company
Communicating Knowledge 96
98. 4 Dimensions of Semantic World
Scalability
Performace
Data
Dynamics
Expressivity
Communicating Knowledge 97
99. Current State of the Art of Technology
Scalability
UbiComp Year Performance
Telco
• 500M triples
Social Net
2005
Enterprise • OWL DLP
Search
Medical • 30B triples
2010
• OWL DL Horst
Expressivity
Communicating Knowledge 98
100. Current State of the Art of Technology
Communicating Knowledge 99
101. Current State of the Art of Technology
Scalability
UbiComp
Year Performance
Telco
• 500M triples
Social Net
2005
Enterprise
• 1~40S (LUBM1000)
Search
• 30B triples
Medical 2010
• 0.01~5S (LUBM1000)
Performance
Communicating Knowledge 100
102. Current State of the Art of Technology
Performance
UbiComp
Telco
Year Performance
Search
• 1~50S (LUBM1000)
2005
• OWL DLP
Social
Net
Medical
• 0.01~5S (L1000)
2009
• OWL DL Horst
Expressivity
Communicating Knowledge 101
103. How to move Maginot Lines?
Scalability
Scalability
?
Expressivity Expressivity
Communicating Knowledge 102
104. 6 Solutions
Current Improved
State of the Art Results
Enhanced algorithm
Materialization
Distributed Computing
Approximation
Lean KR model
Query optimization
+ Query/Data Cache
Communicating Knowledge 103
105. Wining Strategies
Medical Algorithm Materialization
E. Search
Social Net
Mobile
Ubiquitous
Query
Optimization Distribution
(+ Cache)
Lean KR model Approximation
Communicating Knowledge 104