PROPEL . Austrian's Roadmap for Enterprise Linked Data

Editor's Notes

• #8: Projektpartner Kofinanzierung
• #9: To connect data from different sources With the term enterprise linked data we mean to link data from different sources in a business context within an enterprise; this data can be open or closed for external stakeholders. E.g. link different information systems such as enterprise resource planning, customer relationship managment, supply chain management, emails, the web, social media, other sources, etc. As a basis for innovative products and services, to increase efficiency and productivity, and to drive business.
• #11: 1: The characterisation of industries/domains according to defined criteria in order to identify the industries with highest potential for the adoption of the (E)LD concept. 2: An investigation into the data and information management challenges that industries are currently facing, and their formulation as use cases.
• #12: 3: Analysis of the Linked Data community, its current research and development activities as well as open challenges in regards to LD technologies and standards. 4: The development of an integrated roadmap based on the industry, market, and Linked Data community analysis.
• #13: Target private sector enterprises of differnet sizes, the research community and policy makers
• #15: In this phase of the project, we deliberately did not look into particular practical use cases in various industries, but aimed to assess the general suspectability of various sectors to the Linked Data paradigm. The goal was to characterize different sectors along a set of dimensions that indicate how well their structural and technological profile aligns with characteristics of the Linked Data paradigm. In later stages of the project, we then narrowed our analysis to particularly promising application domains and use cases, but in this phase the goal was to take a broad high-level perspective Alignment between Linked Data paradigm and industry characteristics: - Linked Data creates a semantically explicit global information space and is therefore useful for industries that are highly internationalized, networked, knowledge-intense, data-driven etc. Based on such rationales, we developed a set of working hypotheses that I'll explain in a minute Methods: To develop these working hypotheses, we relied extensively on literature research The individual industry characterizations were also mostly informed by desk research, but also through statistical data on industry characteristics (such as R&D intensity, IT spending)
• #16: Explain rationale: Internationalization → The original goal of Linked Data was to create a global information space -> more useful in industries with geographic dispersion Knowledge-intensity: The Semantic web has a strong tradition in knowledge representation, and of course industries where knowledge is important should be more susceptible towards adopting technologies like linked data that help them to manage it Operational complexity: coordination of many activities, cooperative processes, interactions etc. → need for common understanding and a joint infrastructure Network: inter-organizational information flows, need to exchange data Openness: is a core value in the LOD community; a starting point of this project was that these technologies can be applied in a not fully open environment Still, industries that are characterized Linked Data characteristics: Decentralization Linking, sharing and reuse Self-descriptiveness Flexibility and extensibility Openness
• #21: Networked: - information flows across organizational boundaries - need to share and integrate information within and between organizations - Within and across industries Data infrastructure
• #24: The overall economic development influences the IT investments in Austria, slightly positive outlook
• #25: The overall economic development influences the IT investments in Austria, slightly positive outlook
• #26: For 84% of respondents efforts for data and information mng are rather big or even very big. 8% is overlaoded by their efforts.
• #27: The biggest challenges in data management are the cooperation between IT-departments and Lines of Business (LOBs), inconsistency in the business terminology, immature technology, and low data quality.
• #28: Future perspectives for growth are to integrate data from different sources, to create consistency between data and eliminate duplication, and to track communication with customers along different streams and channels (e.g. CRM, e-mail, social media, etc.).
• #30: From the interviews we derived approximately 60 user stories and mapped them to the 18 Foundations listed earlier https://docs.google.com/spreadsheets/d/1x3s0r8Wlg5rt9paa5CwoP0MYwHPqlLj66WbClctbV9w/edit#gid=0 I selected 4 and added them here, however there are others incase you need them
• #31: Result of a stakeholder workshop we conducted with those people… These are the foundations that we use for mapping the user stories from WP2 to the topics from WP3, therefore it is good to introduce them here Analytics Computational linguistics & NLP Concept tagging & annotation (******Personally I’m not convinced about this one******) Data integration Data management Dynamic data / streaming Extraction, data mining, text mining, entity extraction (******Personally I’m not convinced the overlap with NLP******) Logic, formal languages & reasoning Human-Computer Interaction & visualization Knowledge representation Machine learning Ontology/thesaurus/taxonomy management Quality Recommendations Robustness, scalability, optimization and performance Searching, browsing & exploration Security and privacy System engineering
• #32: A not-yet-very-scientific approach… still hope this is interesting, maybe a bit controversial to discuss here!
• #34: Outlines quite clearly what they thought back then the Semantic Web should be…
• #35: Terms to do with Agents and Web Services from conference/journal dictionary – there is no high level foundation so we might need to merge some terms ontologies   ontology management   ontology engineering   ontology languages agents   web services   software agents   services   agent-based … agents research definitly not going up, our community has largely been dominated by ontologies. Might need data before 2006, semanttic Web services topic was already on the decline in 2006.
• #36: The year of ”Linked Data” DBPedia A lot of company use cases that have used SW mentioned:
• #38: Top 10 Companies plot from Sponsor Dictionary Can look over companies per year FORM_
• #45: We identified a number threats for the development of an Enterprise Linked Data ecosystem in the Austrian environment that we have limited control over. LEGAL AND Policy:Inconsistent legal standards across the EU can be a major stumbling block for the development of Linked Data Ecosystems. Here, we are somewhat dependent upon policy-making at the European level. STANDARDISATION:standardization bodies, most notably of course the W3C, may focus their efforts elsewhere. So far, enterprise-related topics have not ranked particularly high in the priorities. TECHNOLOGICAL INNOVATION:Broader developments in the research domain, such as the long-term demographics of the semantic web research community, i.e., the number of suitably qualified science and engineering students entering postgraduate studies. FUNDING:And of course, we as researchers are always afraid of funding cuts and changes in funding priorities .
• #46: AWARENESS: In a sense, the main weaknesses of the Austrian Enterprise Linked Data space is that it does not really exist This can be attributed to a lack of awareness and education As this project shows, there is a community in both academia and industry, but both we as researchers and the industry partners in the project are primarily active internationally This is quite natural in the scientific domain and can be considered a strength for companies like SWC whose client base is mostly international, but not having a "home market" to address is also a weakness. Other weaknesses include the fact that the relevant industry is not headquartered in Austria. In terms of funding, there are high access barriers for funding of research projects And difficulty to fund industry-wide action and adoption at an international level
• #47: Now, we also found that there are significant strenghts. AWARENESS: Strengths include the small, but very active community. This is reflected in the attendance of meetups such as this one. STANDARDIZATION: In terms of standardization, Austria is I THINK overproportionally represented in W3C groups and bodies, which we consider another strength. TECHNOLOGY AND RESEARCH: In terms of technology and research, I think it is fair to say that we are fairly well positioned and that technological innovation in research and practice is a key strength. FUNDING: Funding is of course always an issue to be concerned about, but overall Austria is well-positioned to fund basic and applied research related to Linked Data. Applied research also fosters collaborations and knowledge transfer between universities and industry. LEGAL AND POLICY: ?????????????????????????????????????????????????????
• #48: A major opportunity is the know-how concentrated available at universities and research institutions. Standardization: - certification scheme: interoperability of tools, vendors etc. (e.g., SPARQL) We also consider the EU general data protection regulation with ist strong implications for Linked Data not just a challenge, but an opportunity for the European data infrastructure market in general and the Austrian market in particular
• #50: In the short term, we recommend measures to ensure the visibility of Austrian Linked Data initiatives in an international context
• #51: that are necessary to propel the potential of Linked Data in enterprises.   
• #52: Foster multidisciplinary teams: to make sure the technical research is informed by societal and business requirements To drive technological innovation, develop centers of excellence Goals: Provide a platform for Austrians to showcase technological results from nationally and internationally funded projects Support the setup and development of centers, starting with Linked Data and moving beyond to Big Data Analytics and the Internet of Things
• #53: Flagship projects on ELD legal clearing centers: - demonstrators based on national use cases - broaden findings and solutions to a transnational blueprint - couple outcomes of the flagship projects with the standardization efforts
• #57: Ability to respond to disruptive threats Agile in terms of data modeling and integration proven successful in data and information management The redevelopment of companies coupled with disruptive developments, present companies with the opportunity to adopt ELD, facilitating new business narratives, new processes and new products.
• #58: Given that ELD has its strength as an underpinning technology for new business narratives, Community has also been actively involved in the development of tools (ontologies, RDF, NLP) that
• #59: There is a strong history of fundamental theoretic research in logic and knowledge representation. Important, but if we want the technologies we develop to be relevant, we need to align research priorities with practical needs More customer involvement in business processes and (open) innovation needs to be improved to foster adoption
• #60: In existing environments  While, well established industries have their own home grown IT environments, ELD does not have this level of specialisation and thus has to be adapted beforehand. need to be overcome. For instance, it took a long time to convince industry to use SQL for data management. To the same issues which are still a weakness of ELD. extent there is now resistance against “non-relational” models like Linked Data, where new standards and new skills (like SPARQL) are required.
• #62: This includes centres of excellence
• #66: Turning web of documents into a Web of Data (or: web of things in the world, described by data on the Web) Create links between data from different sources. Traditionally, data published on the Web has been made available as raw dumps in formats such as CSV or XML, or marked up as HTML tables, sacrificing much of its structure and semantics In the conventional hypertext Web, the nature of the relationships between two linked documents is implicit Huge decentralized knowledge base of machine-accessible data Uniquely identifying web objects (documents, images, named-entities, facts, …) Enabling the discovery & interlinking of web objects through semantic metadata Open access to data
• #67: RDF (subject-predicate-object triples) Subjects and object of a triple are URIs that each identify a resource (object may also be a string literal) Predicate specifies how the subject and object are related, also represented by a URI
• #68: Still good practice to use established terminology Term mapping, ontology alignment Mixture of using common vocabularies together with data source-specific terms that are connected by mappings as deemed necessary → Parallel use of arbitrary (self-describing) vocabularies → different URIs for same entities, resolved via sameAs links
• #71: DBPedia: One of the first and most prominent nodes on the LOD cloud Community effort to extract structured (“infobox”) information from Wikipedia provide SPARQL endpoint to the dataset Interlink the Dbpedia dataset with other datasets on the web
• #75: Certain data sets such as DBPedia or Geo Names Database of 10+ Mio geographical names in various languages serve as linking hubs Wikidata Rather than extracting data from Wikipedia, Wikidata is a user curated source for structured information which is included in Wikipedia Includes more than 38 Billion facts expressed as triples (we’ll get to that) from over 650.000 datasets.
• #79: More pragmatic bottom-up approach Ontologies still important for data integration and particular applications, but the idea of agents using ontologies to reason about the global knowledge autonomously had to be given up for now. We will discuss all that in more detail, but for now you can think of Linked Data as a lightweight, somewhat pragmatic, bottom-up approach as opposed to the grand vision that the Semantic Web is or was.
• #84: Basic recipe for publishing and connecting data using the infrastructure of the Web while adhering to its architecture and standards.
• #85: Semantic Web Stack defined as a layer model by the W3C Every layer can access the functionality of the layers below Every layer extends the functionality
• #87: Semantic web: one of the reasons the semantic web didn’t come to be as envisioned was that no global schema exists and people cannot be expected to agree on terms completely Linked Data Paradigm: - You don’t need to agree on all the terms - 'term cherry-picking' approach - bottom-up
• #88: Very widely used..