Steps towards a Data Value Chain

©w Cwowp.ysrtiig-ihntn 2s0b1r2u c k S.aTtI INNSBRUCK www.sti-innsbruck.at
Steps towards a Data Value Chain
José M. García, University of Innsbruck
Tirrenia(Pisa), Italy, June 2013

www.sti-innsbruck.at
Contents
1.
Big Data
2.
Public Open Data
3.
Linked (Open) Data
4.
Data Economy
2

BIG Data

What is Big Data?
• Every day, we create 2.5 quintillion* bytes of data — so much that 90%
of the data in the world today has been created in the last two years alone.
• These data come from everywhere: sensors used to gather climate
information, posts to social media sites, digital pictures and videos,
purchase transaction records, and cell phone GPS signals to name a few.
• These data are big data.**
30
* 10
** http://www-01.ibm.com/software/data/bigdata
4

What is Big Data?
•
“Big data”is a loosely-defined term
•
used to describe data sets so large and complex that they become awkward to work with using on-hand database management tools.
–
White, Tom. Hadoop: The Definitive Guide. 2009. 1st Edition. O'Reilly Media. Pg 3.
–
MIKE2.0, Big Data Definition http://mike2.openmethodology.org/wiki/Big_Data_Definition
Infromation Explosion in data
and real world events (IBM)
5

Big Data Application Areas
Picture taken from http://www-01.ibm.com/software/data/bigdata/industry.html
6

Use case : Climate Research
•
Eiscat and Eiscat 3D are multimillion reserch projects doing environmental research as well as evaluation of the built infrastructures.
–
Observation of climate: sun, troposphere, etc.
–
Simulations, e.g. Creation of artificial Nothern light
–
Run by European Incoherent Scatter Association
•
1,5 Petabytes of data are generated daily (1,5 Million Gigabytes).
–
Processing of this data would require1K petaFLOPSperformance
–
Or1 billionEuro electricitycostsp.a.
7

Large Scale Reasoning
•
Performing deductive inference with a given set of axioms at the Web scale is practically impossible
–
Too manyRDFtriples to process
–
Too much processing power is needed
–
Too much time is needed
•
LarKCaimed at contributing to an ‘infinitely scalable’ Semantic Web reasoning platform by
–
Giving up on 100% correctness and completeness (trading quality for size)
–
Include heuristic search and logic reasoning into a new process
–
Massive parallelization (cluster computing)
8

Large Scale Reasoning
9

Volumes of Data Exceed the Availale Storage Volume Globally
There is a need
to throw the data
away due to
the limited storage
space.
10

Data Stream Processing for Big Data
•
Before throwing the data away some processing can be done at run- time
–
Processing streams of data as they happen
•
Embracing the streaming model
–
Data is seen as a constant flow (sequence) of transient elements
–
Fits naturally with many application domains (sensors, social media, etc.)
•
“Big data” is bringing an inherent set of complexities
–
Data structures exceeding the available memory
–
Approximate/incomplete results are taken as granted
•
Always look at the latest part of a dataset
11

Data Stream Processing for Big Data
•
Logical reasoning in real time on multiple, heterogeneous, gigantic and inevitably noisy data streams in order to support the decision process…
--S. Ceri,E. Della Valle,F. van HarmelenandH. Stuckenschmidt, 2010
window
Extremely large
input streams
streams of answer
Registered Continuous Query
Picture taken from EmanueleDella Valle “Challenges, Approaches, and Solutions in Stream Reasoning”, Semantic Days 2012
Query engine
takes stream subsets for query answering
12

Conclusions
•
Big Data describes datasets so large and complex that they become awkward to work with using on-hand database management tools.
•
Big data application domains are diverse
–
Embracing a big data processing strategy can have a significant impact
•
Tacking the issues of big data processing requires to loose the requirements on completeness and precision
•
Big data on Web scale suffers from an inherent heterogeneity and different levels of expressiveness
•
Complexity is more than just size!
13

Public Open Data
Open
Data

Public Open Data: What is Open Data?
Definitions:
•
Open data is non-personally identifiable data produced in the course of an organisation’sordinary business, which has been released under an unrestricted licence(like the Open Government Licence).
•Open public data is underpinned by the philosophy that data generated or collected by organisationsin the public sector should belong to the taxpayers, wherever financially feasible and where releasing it won’t violate any laws or rights to privacy (either for citizens or government staff).
[linkedgovproject]
http://linkedgov.org
15

Public Open Data: What is Open Data?
Definitions:
The idea behind open data is that information held by government should be freely available to use and re-mix by the public. It’s a movement to make non-personal data:
•open so that it can be turned into useful applications
•support transparency and accountability
•make sharing data between public sector partners more efficient.
The Government is committed to making much more public data openly available. On 22 March 2010, the Prime Minister announced that the Government was going to:
“...use digital technology to open up data with the aim of providing every citizen in Britain with true ownership and accountability over the services they demand from government.”
http://www.idea.gov.uk/
16

Public Open Data: Features of Open Data
Open Data principles [1]:
1.
completeness–all data that can be open (w.r.t. privacy and security) should be open
2.
primary source–all open data should be gathered at their source in raw format
3.
temporal closeness–all open data should be up-to-date
4.
easy access–all open data should be easily accessible
5.
machine readability–all open data should be structured for machine processing
[1] Source [Kaltenböck M., Thurner T., (Hg.): Open Government Data Weißbuch, 2011]
17

Public Open Data: Features of Open Data
Open Data principles [1]:
6.
non-discriminating–all open data should be accessible for everyone
7.
open standards–all open data should use open standards
8.
liberal licensing–all open data should use a liberal licensing without huge obligations for potential users
9.
durability–all open data should be available on a long term basis
10.
non-discriminating usage costs–some open data might involve usage costs. These should be kept as low as possible.
[1] Source [Kaltenböck M., Thurner T., (Hg.): Open Government Data Weißbuch, 2011]
18

Public Open Data: Anditworks(SeeUK)
19

Public Open Data: Anditworks(SeeUK)
•
See UK uses data that have been sourced from data.gov.uk and processed into Linked Data .
•
All the datasets are enriched and cross-linked to additional sources.
•
The visualisationprovides a view centredon a chosen region of the specified size, and most noticeably gives a "pie- chart" that shows the viewer how that region compares with similar regions around it. 20

Public Open Data: Anditworks(police.uk)
21

Public Open Data: Anditworks(police.uk)
•
Different appssuch as„VehicleCrime & Road AccidentMap“, „Crime Sounds“ and„UK Crimeview“ areprovided
•
The usercangeta quick ideaaboutdifferent areasofcitiesandtownsandtheircrimestatistics.
22

Public Open Data
•
Openess: Open Data is about changing behaviour
•
Heterogenity: Different vocabularies are used
•
Interlinkage: Need to link these data sets to prevent data silos
•
LinkedOpen Data
23

Linked Open Data

Motivation: From a Web of Documents to a Web of
Data
•
Web of Documents
•
Fundamental elements:
1.Names(URIs)
2.Documents(Resources) described by HTML, XML, etc.
3.Interactionsvia HTTP
4.(Hyper)Linksbetween documents or anchors in these documents
•Shortcomings:
–Untyped links
–Web search engines fail on complex queries
“Documents”
Hyperlinks
25

Data
•
Web of Documents
•
Web of Data
“Documents”
“Things”
Hyperlinks
Typed Links
26

Data
•
Characteristics:
–Links between arbitrary things (e.g., persons, locations, events, buildings)
–Structure of data on Web pages is made explicit
–Things described on Web pages are named and get URIs
–Links between things are made explicit and are typed
•
Web of Data
“Things”
Typed Links
27

Google Knowledge Graph
•
“A huge knowledge graph of interconnected entities and their attributes”.
AmitSinghal, Senior Vice President at Google
•
“A knowledge based used by Google to enhance its search engine’s results with semantic-search information gathered from a wide variety of sources”
http://en.wikipedia.org/wiki/Knowledge_Graph
•
Based on information derived from many sources including Freebase, CIA World Factbook, Wikipedia
•
Contains about 3.5 billion facts about 500 million objects
28

Google Knowledge Graph
http://goo.gl/zp3IH
29

Linked Data –a definition and principles
•
Linked Datais about the use of Semantic Web technologies to publish structured data on the Web and set links between data sources.
Figure from C. Bizer
30

Linked Data –a definition and principles
1.
Use URIs as names for things.
2.
Use HTTP URIs so that people can look up (dereference) those names.
3.
When someone looks up a URI, provide useful information, using the standards (RDF*, SPARQL)
4.
Include links to other URIs. so that they can discover more things.
31

5-star Linked OPEN Data
★Available on the web (whatever format) but with an open licence, to be Open Data
★★Available as machine- readable structured data (e.g. excel instead of image scan of a table)
★★★as (2) plus non-proprietary format (e.g. CSV instead of excel)
★★★★All the above plus, Use open standards from W3C (URIs, RDF and SPARQL) to identify things, so that people can point at your stuff
★★★★★All the above, plus: Link your data to other people’s data to provide context
32

★Available on the web (+ open licence)
•
Easy to publish web data
•
Data can be easily accessed and stored locally
•
Data can be entered manually into another system
Source [Bauer F., Kaltenböck, M.: Linked Open Data: The Essentials, 2012]
33

★★Available as structured data
•
All benefits from ★
•
Data can be directly processed with proprietary software
•
Easy to export it into another structured format
34

★★★Non-proprietary format is used
•
All benefits from ★★
•
No need to pay for a format controlled by a single organization
35

★★★★Use open standards to identify things
•
All benefits from ★★★
•
Link to data from anywhere, either on the web or locally
•
It can be bookmarked and parts of the data can be reused
•
Access to data items can be optimized (caching, load balancing, etc.)
•
BUT the publisher needs to identify separable items, assign URIs to each one, and
36

★★★★★Data is linked to provide context
•
All benefits from ★★★★
•
New data of interest can be discovered while consuming other
•
Data schema can be obtained
•
Added value to the data
•
Linked datasets are discoverable
•
BUT resources have to be invested to link datasets Source [Bauer F., Kaltenböck, M.: Linked Open Data: The Essentials, 2012]
37

LOD Cloud May 2007
Figure from http://linkeddata.org/
38

LOD Cloud May 2007
Basics:
The Linked Open Data cloud is an interconnected set of datasets all of which were published and interlinked following the Linked Data principles.
Facts:
•Focal points:
•DBPedia: RDFized vesion of Wikipiedia; many ingoing and outgoing links
•Music-related datasets
•Big datasets include FOAF, US Census data
•Size approx. 1 billion triples, 250k links
39

LOD Cloud March 2009
40

LOD Cloud September 2011
41

LOD Cloud September 2011
Facts:
•295 data sets
•Over 31 billion triples
•Over 504 billion RDF links between data sources
42

Linked Open Data –silver bullet for data integration
•
Linked Open Data can be seen as a global data integration platform
–
Heterogeneous data items from different data sets are linked to each other following the Linked Data principles
–
Widely deployed vocabularies (e.g. FOAF) provide the predicates to specify links between data items
•
Data integration with LOD requires:
1.
Access to Linked Data
•
HTTP, SPARQL endpoints, RDF dumps
•
Crawling and caching
2.
Normalize vocabularies –data sets that overlap in content use different vocabularies
•
Use schema mapping techniques based on rules (e.g. RIF, SWRL) or query languages (e.g. SPARQL Construct, etc.)
3.
Resolve identifies –data sets that overlap in content use different URIs for the same real world entities
•
Use manual merging or approaches such as SILK (part of Linked Data Integration Framework) or LIMES
4.
Filter data
•
Use SIVE ((part of Linked Data Integration Framework)
See:
http://www4.wiwiss.fu-berlin.de/bizer/ldif/
43

Example -Mashup: DBPedia Mobile
•
Geospatial entry point into the Web of Data.
•
It exploits information coming from DBpedia, Revyu and Flickr data.
•
It provides a way to explore maps of cities and gives pointers to more information which can be explored
44
Try yourself: http://wiki.dbpedia.org/DBpediaMobile
Pictures from DBPedia Mobile
44

From Intranet to Enterprise Data Web around a knowledge hub

Data Economy
“Your data is worth more if you give it away.”
Commission Vice President
NeelieKroes
“Your data is worth more if you give it away.”
Commission Vice President
NeelieKroes

What is Data Economy?
•
Non tangible assets (i.e. data) play a significant role in the creation of economic value
•
Data is nowadays more important than, for example, search or advertisement
•
The value of the data, its potential to be used to create new products and services, is more important than the data itself
47

Why a Data Economy?
•
Total market for public sector information €28 billion in 2008for the EU27
•
Annual growth of 7% leading to around €32billion in 2010
•
Estimated €40 billion annual boost for the European economy.
•
The total direct and indirect economic gains across the whole EU27 economy would be in the order of €140 billionannually.
See:
Review of recent studies on PSI re-use and related market developments, G. Vickery, August 2011.
48

Why a Data Economy?
•
New businesses can be built on the back of these data: Data are an essential raw material for a wide range of new information products and services which build on new possibilities to analyseand visualisedata from different sources. Facilitating re-use of these raw data will create jobs and thus stimulate growth.
•
More Transparency: Open data is a powerful tool to increase the transparency of public administration, improving the visibility of previously inaccessible information, informing citizens and business about policies, public spending and outcomes.
•
Evidence-based policy making and administrative efficiency: The availability of solid EU-wide public data will lead to better evidence- based policy making at all levels of government, resulting in better public services and more efficient public spending.
See:
http://europa.eu/rapid/pressReleasesAction.do?reference=MEMO11/891&format=HTML&aged=0&language=EN&guiLanguage=en
49

CombiningOpen Data andServices –Tourist MapAustria
•
Use LOD to integrate and lookup data about
–
places and routes
–
time-tables for public transport
–
hiking trails
–
ski slopes
–
points-of-interest
50

LOD data sets
•
Open Streetmap
•
Google Places
•
Databases of government
–
TIRIS
–
DVT
•
Tourism & Ticketing association
•
IVB (busses and trams)
•
OEBB (trains)
•
Ärztekammer
•
Supermarket chains: listing of products
•
Hofer and similar: weekly offers
•
ASFINAG: Traffic/Congestion data
•
Herold(yellow pages)
•
City archive
•
Museums/Zoo
•
News sources like TT (Tyrol's major daily newspaper)
•
StatistikAustria
•
Innsbruck Airport (travel times, airline schedules)
•
ZAMG (Weather)
•
University of Innsbruck (Curricula, student statistics, study possibilities)
•
IKB (electricity, water consumption)
•
Entertainment facilities (Stadtcafe, Cinema...)
•
Special offers (Groupon)

•
Data and services from destination sites integrated for recommendation and booking of
–
Hotels
–
Restaurants
–
Cultural and entertainment events
–
Sightseeing
–
Shops
52

•
Web scraping integration
•
Create wrappers for current web sites and extract data automatically
•
Many Web scraping tools available on the market
53

•
Integration intoa comprehensivemapofmulti-channelcommunication, seekdabookingengine, LinkedOpen Data andon theflyserviceintegrationasyoupayto generate addedvalueforbusinessesaswell ascustomers
•
Combinationofmultichannelcommunicationandyieldmanagement
–
SemanticCommunication Engine Innsbruck (SCEI)
–
seekdabookingsolutions
•
enrichedwithLinked(Open) Data
–
Machineunderstandableinterlinkeddata
–
Bike andhikingtrails, sightinformation, etc.
•
and on the fly service integration as you pay
–
Solutions for ad-hoc service integration for touristic destination sites
–
Bike rental, ski passes, etc.
–
Services are quickly integrated through scraping (and later through legal frameworks and backendintegration in case of business volumes)

• Based on Open
Street Map
55

• Based on Open
Street Map
• Increase on-line
visibility for hotels
and destinations via
multi-channel
communication –
SCEI
SCEI
56

• Based on Open
Street Map
multi-channel
communication –
SCEI
• Hotels, ski passes,
etc. are directly
bookable – seekda
engine
SCEI
57

• Based on Open
Street Map
multi-channel
communication –
SCEI
etc. are directly
bookable – seekda
engine
• LOD to integrate
and lookup data
about hiking trails,
ski slopes, etc.
LOD
SCEI
58

• Based on Open
Street Map
multi-channel
communication –
SCEI
etc. are directly
bookable – seekda
engine
• LOD to integrate
and lookup data
about hiking trails,
ski slopes, etc.
• On the fly service
integration as you
pay
LOD
SCEI
59

“There's No Money in Linked (Open) Data”
http://knoesis.wright.edu/faculty/pascal/pub/nomoneylod.pdf
•
It turns out that using LOD datasets in realistic settings is not always easy.
–
Surprisingly, in many cases the underlying issues are not technical but legal barriers erected by the LD data publishers.
–
Generally, mostly non-technical but socio-economical barriers hamper the reuse of date (do patents and IPR protections hamper or facilitate knowledge reuse?).
–
Business intelligence
–
Dynamic Data
–
On the fly generation of data
60

Thanks for your attention
61

Steps towards a Data Value Chain

More Related Content

Steps towards a Data Value Chain