Embed Size (px)
Citation preview
Austria's Roadmap for Enterprise Linked Data
14:45 … Hot Drinks 15:00 … Welcome 15:10 … The Project 15:20 … The Findings 15:35 … Conclusions, Roadmap 15:45 … Guest Presentation: Data Market Austria 16:00 … Get your Book and hand over to Vienna Open Data MeetUp
3
The Austrian Data Eco System
4
The Austrian Data Eco System
5
The Austrian Data Eco System
5
Sabrina Kirrane (WU, Privacy and Sustainable Computing Lab)Julia Neuschmidt (IDC Austria)Mihai Lupu (Researchstudios Austria)Elmar Kiesling (TU, Linked Data Lab)Thomas Thurner (SWC + School of Data)
The PROPEL ProjectPropelling the Potential of Enterprise Linked Data
15.12.2016
PROPEL 8
Emerging concept for data exchange and integration Based on standard web technologies Shifting away from a predominantly academic perspective, we
conceive Linked Data as a promising disruptive technology for enterprise data management.
Source: blog.backand.com
Linked Data
PROPEL 9
The project goal
Survey industry and market needs, technological challenges, and open research questions on the use of Linked Data in a
business context.
FFG ICT of the Future 2014/2015 Exploratory study Project duration Nov 2015 – Dec 2016 Consortium: IDC Austria, Technical University of Vienna,
University of Economy Vienna, Semantic Web Company
PROPEL 10
Approach
Which industries are the most likely to adopt LD technologies?
What are the key drivers, inhibitors and needs in data management from a demand sideperspective?
PROPEL 11
Approach
What recommendations are necessary for enterprises, policy makers and researchersin order to propel the adoption of LD in enterprises?
What are technological and standardisation opportunities and challenges?
PROPEL 12
Approach
Stakeholder Workshop Interviews Survey respondents
Comprehensive Literature research
Internalworkshops
www.linked-data.at
Findings:Sectoral Analysis of Linked Data Potential
PROPEL 14
Sectoral Analysis of LD Potential
Goal: • Exploratory sectoral assessment of Linked Data adoption potential • Alignment between Linked Data paradigm and industry characteristics• Broad high-level, theoretical perspective
Methods:• Industry classification: NACE rev. 2 top level sections,
with selective use of more detailed classes• Extensive literature research• Analysis of statistical data on industry characteristics
(R&D intensity, ICT spending,..)
• Industry expert interviews• Internal validation survey
15
Working Hypotheses Sectoral Characteristics → Adoption
16
Sectoral Characteristics - Results
17
High potential sectors
✅Highly networked✅Strong (potential) impact
of ICT-based innovation✅Data- and ICT-intense☑Global scope☑Knowledge-intense☑Complex operations☑Relatively open☑Some uptake of web
technologies
18
Medium potential sectors
✅Highly networked✅ICT- and data-intense✅Strong (potential) impact
of ICT-based innovation☑Highly internationalized☑Complex operations❌Have not embraced
openness❌Limited uptake of web
technologies
PROPEL 19
Lower potential sectors
✅Structural characteristics mostly favorable
❌Moderate ICT dynamics❌Have not embraced
openness❌Trailing web technology
uptake
PROPEL 20
Results
Broad potential for ELD across a large spectrum of industries Focus on ”openness” and “web-centric positioning” in academic
discussions may inhibit enterprise adoption
Virtually all sectors in developed economies exhibit structural characteristics that favor LD adoption:• Actors in a highly networked global economy• Increasingly data-driven and knowledge-intense• Cross-organizational operations
However, various sectors• are laggards in the technological dimensions and• have untapped potential for ICT-based innovation
Findings:Market Forces
PROPEL 22
Market Forces Economy:
• Positive economic development in Austria leads to a growth in IT spending and we expect investments solutions for data and information management
Efficiency: • Organizations focus primarily on costs. Data and information management
solutions and LD can have positive effects in terms of transforming businesses, increasing efficiency and driving growth
Digital Transformation: • Data and information management is a key asset for digital transformation, and
concepts around Linked Data can support the transformation process
PROPEL 23
Market Forces Culture:
• Missing innovation culture in some organisations might be inhibitors for the uptake of new technologies
Data driven networked global economy: • Growing need to break up silos, and to share data across organizational
boundaries.
Digital life of citizens: • High Internet adoption and user demands for new digital products and services
lead to redefinition and expansion of services.
PROPEL 24
Market Forces Technology:
• New technologies like cloud, big data, IoT and cognitive computing/machine learning change the way our data is managed.
Data security and privacy: • Common barriers to adoption of new technology; at the same time security
concerns provide an opportunity for solution providers to generate revenue out of their security solutions and services.
Regulations: • General Data Protection Regulation forces organizations to take a fresh look on
how they manage their data.
PROPEL 25
Big efforts for data and information management
Demand-side analysis
PROPEL 26
Demand-side analysis
PROPEL 27
Demand-side analysis
Findings:Technology
PROPEL 29
Interviews
23 interviews:
Domains Consulting, Engineering, Environment, Finance and Insurance,
Government, Healthcare, ICT, IT, Media, Pharmaceutical, Professional Services, Real Estate, Research, Startup, Tourism, Transports & Logistics
Roles Business Intelligence, CEO, Chief Engineer, Data and Systems Architect,
Data Scientist, Director Information Management, Enterprise Architect, Founder, General Secretary, Governance, Risk & Compliance Manager, Head of Communications and Media, Head of Development, Head of HR, Head of R&D, Innovation Manager, Information Architect, IT Project Manager, Management, Managing director, Marketing Analyst, Principle System Analyst, Project Coordinator, Researcher, Technical Specialist
Note: Instead of explaining them what ELD is, we gathered their
technology/research expectations from a more general SW perspective
PROPEL 30
Technologies in need…
Analytics Computational linguistics & NLP
Concept tagging & annotation Data integration
Data management Dynamic data / streaming
Extraction, data mining, text mining,
entity extraction
Logic, formal languages &
reasoning
Human-Computer Interaction & visualization
Knowledge representation Machine learning
Ontology/thesaurus/taxonomy
management
Quality & Provenance Recommendations
Robustness, scalability,
optimization and performance
Searching, browsing & exploration
Security and privacy System engineering
PROPEL 31
Monitoring SW community major venues:• ISWC (since 2006), ESWC (since 2006), SEMANTiCS
(since 2007), JWS (since 2006), SWJ (since 2010)
3 seminal papers:
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Community Analysis
PROPEL 32
Topic Categorisation
PROPEL 33
Semantic Web/Linked Data over time…
Subtopics:
Expressing Meaning
Knowledge Representation Ontologies
Agents
Evolution of Knowledge
PROPEL 34
Knowledge Representation & Reasoning
PROPEL 35
Semantic Web/Linked Data over time…
Early adopters: MITRE Chevron British Telecom Boeing Ordnance Survey Eli Lily Pfizer Agfa Food and Drug Administration National Institutes of Health
Software adopters/products: Oracle Adobe Altova OpenLink TopQuadrant Software AG Aduna Software Protége SAPHIRE
PROPEL 36
LD Adopters - Companies
PROPEL 37
LD Adopters - Companies
PROPEL 38
PROPEL 39
Semantic Web/Linked Data over time…
The authors claim that "early research has transitioned into these larger, more applied systems, today’s Semantic Web research is changing: It builds on the earlier foundations but it has generated a more diverse set of pursuits”.
PROPEL 40
Looking to the future
ROADMAP
PROPEL 42
Roadmap Formulation
1. Austrian perspective SWOT analysis:1. Awareness and education2. Technological Innovation and Research3. Standardization4. Legal and Policy5. Funding
2. Development of prioritized recommendations
43
SWOT
44
Know the threats
45
See the weaknesses
46
Build on strength
47
Take the opportunities
Activities
54
Long-termSupport
emerging Linked Enterprise Data
ecosystems
Establish centers of excellence
Position Austria as a hotspot for LED research
and innovation
Awareness and Education
Legal and Policy
FundingTechnological Innovation Research
Medium-termDevelop key foundational technologies
Institutional and technological focus on key issues and domains
Short-termCluster
stakeholders and efforts
Get momentum from new
funding lines
Supporting studies and pilot
projects
Take-home messages
PROPEL 55
PROPEL 56
"Use the power of ELD!" Many industries are facing disruptive change Even conservative industries see a need for a "two speed IT" Linked Data can be both a disruptive force and a means to
respond to disruptive change Key ELD technologies are mature and have been successfully
applied in many domains Linked Data is agile and flexible ELD is a enabler for product, process and business model
innovation!
PROPEL 57
"ELD is the backbone for the developing content industry"
Linked Data is particularly relevant for online businesses
(media, e-commerce, etc.)
ELD provides a platform to generate and leverage economic network effects typical for these industries
Tools to enrich digital products and make them interchangeable within a broader digital environment
PROPEL 58
"We need to align research priorities and practical needs"
Continued fundamental basic research necessary, but:
Industry needs should be reflected in applied research agendas
More courage to apply cutting-edge technologies in industry needed!
PROPEL 59
"ELD has to convince stakeholders to embrace change"
Technological, behavioural and cultural adoption barriers New skill sets required
To instigate change, ELD must ..make sense from a business perspective
→ clear business cases, fast returns, tangible, quantifiable benefits
..lower entry barriers• by playing well with existing infrastructure
• through open source/freemium/experimental models
..address security, privacy, and compliance concerns
PROPEL 61
"Need to support [and subsidize] emerging ELD ecosystems" Prototypical example of a technology with strong economic network effects Flagship implementations and pioneering projects are key to furthering the
growth of ELD in Austria. Both financial and infrastructural support are necessary in order to
accelerate the development of the sector.
Core preparatory steps include:
• Base infrastructures (stores, services, data) to build solutions on top
• Project related funding
Backup
Linked Data in a Nutshell
PROPEL Workshop May 10, 2016
Linked Data from 10,000 foot...
• Best practices for publishing and connecting structured data on the Web
• Goal: Creating a global data space
Hype
rlink
s
Type
d Li
nks
Web of Documents Web of Data
66
... and up-close Graph-based data model that captures statements about
things in the world Subject-predicate-object triples Use of URIs as globally unique identifiers
PROPEL
http://example.com/alice
http://xmlns.com/foaf/0.1/knows
http://example.com/bob
:alice
foaf:knows
:bob
PROPEL 67
PrinciplesAnyone can…
• publish data• create URIs • choose or create vocabularies to represent their data• refer to Linked Data published by others
Result:• Decentralized data infrastructure (> 650.000 datasets)• Machine-readable, and -discoverable data sets• Bottom-up "pay as you go" data integration
PROPEL 68
Key ideas
Explicit SemanticsWeb of data Graph-based
Network effects
Global data spaceBottom-up
FlexibleAgile Machine readable
Interoperable
Ad-hoc integration
Linking
Decentralized InferenceDiscovery
a
b
cx
y
Emergent
Open
Evolution of the Linked Data Cloud: 2007http://lod-cloud.net
Evolution of the Linked Data Cloud: 2008http://lod-cloud.net
Evolution of the Linked Data Cloud: 2009http://lod-cloud.net
Evolution of the Linked Data Cloud: 2010http://lod-cloud.net
Evolution of the Linked Data Cloud: 2011http://lod-cloud.net
Media
Geographic
Publications
Social Networking
Government
Cross-Domain
LifeSciences
User Generated
ContentLinguistics
Evolution of the Linked Data Cloud: 2014http://lod-cloud.net
PROPEL 75
ELD and LED
Enterprise Linked Data: Internal use of LD technologies within organizations, e.g.,
• to integrate heterogeneous systems at the data level• for advanced content/knowledge/… management • as a basis for innovative products and services
Linked Enterprise Data:• Cross-organizational data integration• Data markets and data ecosystems• Decentralized infrastructure for a networked economy
What's the difference between Linked Data and... ?
PROPEL 76
PROPEL 77
Linked Data vs. Open Data
Overlaps:• Openness is a core principle in the design of LD• Many Linked Data sets published under an open license
→ Linked Open Data and LD often used interchangeably
Key differences:• Linked Data technologies can be used without publishing data –
e.g., for internal and external data integration.• Not all open data will ever be linked (the majority will remain in
formats such as csv, txt etc.)
Linked Data vs. “The” Semantic Web
Overlaps:• "LD is the Semantic Web done right" (Tim Berners-Lee)• Semantic web is made up of Linked Data.• Linked Data is based on Semantic web standards.
Key Differences:• Semantic Web was all about "semantifying" the web, Linked Data is
based on web standards (URIs, http), but doesn't center around web pages.
• LD is a more pragmatic "bottom-up" approach.• "Linked Data is mainly about publishing structured data in RDF using URIs
rather than focusing on the ontological level or inference."
M. Hausenblas "Exploiting Linked Data For Building Web Applications" IEEE Internet Computing, 2009
PROPEL 79
Linked Data vs Big DataOverlaps:
• LD as a whole is big ( *)• No rigid up-front (e.g., relational) data model • Big Data technologies (e.g., Hadoop) are used to handle LD• LD can represent knowledge extracted from big unstructured data
Key Differences:• Individual linked data sets are typically not "big" per se
(e.g., English DBPedia dump currently < 5 GB)• LD is structured and semantically explicit,
"big data lakes" are typically neither• Big data based on distributed data infrastructures within an organization (e.g.,
Hadoop clusters), LD creates a decentralized, globally distributed data infrastructure
http
://lo
dlau
ndro
mat
.org
as
per 2
016-
05-1
0
PROPEL 80
Linked Data vs Knowledge Graphs
Facebook Open Graph Google's knowledge graphExamples:
Linked Data vs Knowledge Graphs
Overlaps:• Knowledge Graphs also represent explicit semantics in a
graph-based data model• Both are often used to facilitate semantic search• Knowledge graphs can use open standards (e.g., RDFa)
Key differences: • Proprietary (data and technologies), closed "ecosystem"• Tightly integrated with services• Typically not published externally → no way to link to
PROPEL 82
ReferencesVideos:
Tim Berners-Lee: The next Web of open, linked data (16:52) Linked Data (and the Web of Data) Manu Sporny: What is Linked Data (12:09) Michael Hausenblas: Quick Linked Data Intro (3:14) Annenberg Networks Theory Seminar with Tim-Berners-Lee Metaweb (now defunct): Words vs entities
Tutorial: Linda Project: Linked Data Primer
Articles: C. Bizer, T. Heath, and T. Berners-Lee. Linked Data - The Story So Far. International Journal on Semantic Web and
Information Systems, 5(3):1 – 22, 2009.
Books: T. Pellegrini, H. Sack, and S. Auer, Eds., Linked Enterprise Data. Heidelberg: Springer Berlin, 2014. Tom Heath, Christian Bizer (2011). Linked Data - Evolving the Web into a Global Data Space.
Morgan & Claypool, 2011. EUCLID Project Consortium (2014). Using Linked Data Effectively. Hitzler, Rudolph, Krötzsch (2009). Foundations of Semantic Web Technologies. Chapman & Hall/CRC
Linked Data Principles
1. Use URIs to identify things2. Use HTTP URIs so that people can look up
those names3. 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
Design Issues: Linked Data notes, Tim
Berners-Lee
The Semantic Web Technology Stack
http://bnode.org/blog/2009/07/08/the-semantic-w
eb-not-a-piece-of-cake
PROPEL 85
Selected Linked Data Standards/Technologies URIs + HTTP:
• Web infrastructure that provides global identifiers for all objects
RDF: • provides a generic graph-based data model for describing things• various serializations
RDFS and OWL• Basis for the definition of vocabularies
(i.e., collections of classes and properties)• Expressed in RDF• Facilitates inference (using reasoning engines)
SPARQL:• Graph pattern-based query language (and protocol) for RDF data
Vocabularies
Many vocabularies beyond those defined in the RDF standard Collections of defined relationships and classes of resources Vocabulary definition and reuse is a key semantic web principle
Adapted from Euclid learning materials by Barry Norton
Best practices:• Terms from well-known vocabularies
should be reused wherever possible• New terms should be defined only if you
can not find required terms in existing vocabularies
• Feel free to mix terms from different vocabularies and to extend the vocabularies with additional terms in your own namespace
Examples of common VocabulariesVocabulary Description Classes and Relationships
Friend-of-a-Friend (FOAF)
Vocabulary for describing people.
foaf:Person, foaf:Agent, foaf:name, foaf:knows, foaf:member
Dublin Core (DC) Defines general metadata attributes.
dc:FileFormat, dc:MediaType, dc:creator, dc:description
Semantically-Interlinked Online Communities (SIOC)
Vocabulary for representing online communities.
sioc:Community, sioc:Forum, sioc:Post, sioc:follows, sioc:topic
Music Ontology (MO) Provides terms for describing artists, albums and tracks.
mo:MusicArtist, mo:MusicGroup, mo:Signal, mo:member, mo:record
Simple Knowledge Organization System (SKOS)
Vocabulary for representing taxonomies and loosely structured knowledge.
skos:Concept, skos:inScheme, skos:definition, skos:example
Adapted from Euclid learning materials by Barry Norton
PROPEL 88
Linked Data from an Application Development Perspective Data is self-describing (applications can dereference
URIs that identify vocabulary terms in order to find their definition)
Use of HTTP as standardized data access mechanism and RDF as a standardized data model simplifies data access compared to Web APIs, which rely on heterogeneous data models and access interfaces
Web of Data is open, i.e., applications do not have to be implemented against a fixed set of data sources, but can discover new data sources at run-time by following RDF links.
