International Symposium of the THESEUS Research Program

Technologies for the Internet of Services

German Federal Ministry for Economy and Technology, Berlin

June 29-30, 2009

EXECUTIVE SESSION KEYNOTES

	Dagmar G. Wöhrl, Parliamentary State Secretary THESEUS – Research Funding for the Benefit of the Society
	Prof. João Schwarz da Silva Network and Communication Technologies, DG-INFSO, European Commission European Research for the Internet of the Future (Download)
	Prof. Miki Haseyama Graduate School of Information Science and Technology, Hokkaido University Information Explosion Era and Information Grand Voyage Project (Download)
	Prof. Lutz Heuser Executive Vice President & Head of SAP Research THESEUS: Technologies for the Internet of Services (Download)
	Pieter van der Linden Technical Director and Coordinator of the QUAERO Program QUAERO: Technologies for automatic Analysis and Classification of Multimedia and Multilingual Documents (Download)
	Martin Kimmich Team Leader "knowledge management methods and processes" at Festo AG & Co. KG Process oriented Knowledge Management (Download)
	Martin Jetter General Manager of IBM Germany Technologies for a Smarter Planet (Download)
	Dion Hinchcliffe Internationally recognized business strategist and enterprise architect Business Perspectives of Web X.0
	Dr. Rainer Ruggaber Project Manager of THESEUS / TEXO The TEXO Service Platform (Download)
	Prof. Hartmut Raffler Head of the division "Information and Communications" within Siemens Corporate Technology The Nessi Open Service Framework (Download)
	Prof. Yanbo Han Professor at the Institute of Computing Technology and the Graduate University of Chinese Academy of Sciences in Beijing, China.Director of the Institute of Service Engineering, Shandong University of Science and Technology in Qingdao, China. The VINCA Service Platform (Download)
	Dr. Márta Nagy-Rothengass managing the "Technologies for Information Management" Unit in Luxembourg, European Commission European ICT research supporting Intelligent Information Management through Semantic Technologies (Download)
	Prof. Rudi Studer Full Professor in Applied Informatics at the University of Karlsruhe, Institute AIFB. In addition, he is director of the Karlsruhe Service Research Institute (KSRI) Semantic Technologies – Achievements and Perspectives (Download)
	Agata Filipowska Teaching and Research Assistant at Poznan University of Economics, Department of Information Systems Semantic Technologies to support Business Process Management (Download)

 

ENTERPRISE PERSPECTIVES

BP1: Economics 2.0; Dion Hinchcliffe, Hinchcliffe & Company

BP2: Business Model Research for the Internet of Services; Thomas Renner & Nico Weiner, Fraunhofer IAO

Under the term "Internet of Services" different innovative IT solutions have been developing which significantly influence markets, organizations and the behaviour of users, e.g. sales-force.com and Amazon. With today's semantic and service-oriented technologies THESEUS promotes the "Internet of Services" as the trend with the greatest impact for our world of tomorrow. A concrete business model is probably the most significant success factor leading to a successful service-offering. This track focalize the business view of the internet of tomorrow and reveals the opportunities of future business models. The topics are theory and practice of service-based and open source business models. Join us to discuss with distinguished experts about future business models on the web.

Business Models In The Internet Of Services - State Of The Art And Trends; Nico Weiner, Fraunhofer IAO
The term "business model" has been used increasingly since the late 90s in relation to the Dot-Com-era. Since the upcoming of the Web 2.0 phenomenon around 2005 and the new service-trends in the web the term gained new popularity. Starting with a short introduction into the scientific landscape we will give a general view on current service-based models and their special characteristics. Additionally the presentation shows trends and characteristics of venture-capital funded start-ups in this area based on a survey done by Fraunhofer IAO.

Dispersing The Clouds On Business Models For The Future Internet - Show Me The Money; Dr. Arian Zwegers, European Commission The emerging Internet of Services will lead to new business models and business strategies that companies will use in future. This presentation will consider well-known business models based on Software-as-a-Service and Cloud Computing, and put them in the context of the Future Internet, both in a technological and socio-economic manner. A number of business strategies will be discussed as well. Finally, the place of Internet business models in the current EU programme for research and technology development will be indicated.

Rethinking Business Models Alexander Osterwalder, Phd. After this talk you will understand why business models and particularly business model innovation are important. We will discuss how they can be made a systematic part of an organization through a tested visual approach to describing, developing and sharing business models. This approach is applied in organizations such as 3M, IBM, Telenor, Ericsson and the Government of Canada. The talk will be accompanied by numerous business model innovation examples ranging from traditional industries to Web-based companies. Participants will be challenged to think out of the box in a number of short exercises. Finally, we will briefly look at how the presenter is publishing a book on business model innovation based on an innovative business model (co-creation with 350+ paying participants on an online platform, disintermediation of publishers, designed & visual format).

The Commercial Open Source Business Model  Dirk Riehle, SAP Labs LLC. Commercial open source software projects are open source software projects that are owned by a single firm that derives a direct and significant revenue stream from the software. Commercial open source at first glance represents an economic paradox: How can a firm earn money if it is making its product available for free as open source? This talk presents the core properties of commercial open source business models and discusses how they work. Using a commercial open source approach, firms can get to market faster with a superior product at lower cost than possible for traditional competitors. The paper shows how these benefits accrue from an engaged and self-supporting user community. Lacking any prior comprehensive reference, this work is based on an analysis of public statements by practitioners of commercial open source. It forges the various anecdotes into a coherent description of revenue generation strategies and relevant business functions.

BP3: Microblogging in Enterprises; Dirk Riehle, SAP Labs LLC

SERVICE INFRASTRUCTURES

SI1: SMILA Tutorial;Igor Novakovic, empolis

Mastering Unstructured Information with SMILA - the SeMantic Information Logistics Architecture

The amount and diversity of information is growing exponentially, mainly in the area of unstructured data, like emails, text files, blogs, images etc. Poor data accessibility, user rights integration and the lack of semantic meta data are constraining factors for building next generation enterprise search and other document centric applications. Missing standards result in proprietary solutions with huge short and long term cost. Overcoming these problems is a key issue for gaining agility in an organisation.

SMILA is an extensible framework for processing unstructured information in the enterprise. Besides providing essential infrastructure components and services, SMILA also delivers ready-to-use add-on components, like connectors to most relevant data sources. Using the framework as their basis will enable developers to concentrate on the creation of higher value solutions, like semantic search applications, information extraction and the like.

SMILA is an open source project under the umbrella of the eclipse foundation. It is also a part of the German research programme THESEUS. Further information can be found at www.eclipse.org/smila and http://theseus-programm.de/en-US/home/default.aspx.

This one day SMILA Tutorial will introduce the concepts and approach behind the framework, how to use it to build an application and how to integrate new components into it. The topics that will be addressed are:

• SMILA in a nutshell
• Installation, crawler and service configuration, and building a search application with SMILA and Lucene
• Creating a simple native SMILA component
• Using Web Services as SMILA components (Open Calais as exercise)
• Presentation of some Demo Applications based on SMILA

Participants should have a basic understanding of JAVA and programming. For the practical exercises, a laptop running Windows or Linux is required. Participants will receive a CD-ROM containing the most recent SMILA release, Eclipse, Protégé and a Java SDK.

The tutorial is presented by Igor Novakovic, Empolis GmbH, Germany

(Download SMILA in a nutshell)

SI2: Service Platforms; Dr. Rainer Ruggaber, SAP AG
Prof. Yanbo Han, Chinese Academy of Services

SI3: Service Innovation

Service Innovation – The TEXO Innovation Process and the TEXO Innovation Repository; Prof. Dr. Krcmar / Christoph Riedl, M.Sc. Technische Universität München

Innovation Mining Cockpit – Finding and Analysing Innovation-Relevant Data on the Web; Dipl.-Inf. Jan Finzen Fraunhofer IAO

Feedback Mechanisms for Service Improvements in the Agora Service Marketplace; Dr. Norman May SAP AG

Topic Detection and its Application to Community Mining; Dr. Alexandru Berlea SAP AG

SEMANTIC TECHNOLOGIES

ST1: Mobile Services; Dennis Boehres, Neofonie GmbH
Daniel Porta & Norbert Reithinger, DFKI GmbH

This workshop is subdivided into three segments. In the first part, Dennis Boehres of Neofonie GmbH talks about mobile frontends and the role they can play in augmenting semantic search services. By integrating and bridging mobile, local and social contexts they show promise in delivering the capabilities of semantic search engines right into the palms and pockets of consumers. Mr. Boehres gives an account of the challenges the company faced during the development of mobile frontends to the company's search engine "WeFind" and demonstrates the iPhone frontend.

In the second part of the tutorial, Daniel Porta and Norbert Reithinger, DFKI GmbH, will demonstrate how the mobile TEXO system for the iPhone is developed. It uses the Ontology Based Dialogue Platform (ODP) developed in the Core technology Cluster (CTC) work package 4 "Situation Aware Dialog Shell" as a mediator between the mobile client and the TEXO platform. We will give a short introduction to the ODP approach and will then sketch how the technology was adapted to realize a mobile, multimodal front-end for the TEXO platform.

In the last part, we will, together with the workshop attendees, do some brainstorming about future mobile solutions for the Internet of Services.

ST2: Visualizing Semantics; Kawa Nazemi, Fraunhofer IGD

Following the areas of computer-based processing of data and information, semantics processing defines the third generation of computer applications. Here, semantics is not only explicitly defined by, e.g., ontologies or topic maps, but also implicitly in large data sets. Semantic technologies and semantic information offer new interaction and visualization possibilities, to enhance and optimize the information-flow between users and computer applications. In the THESEUS research program different aspects of semantic technologies are emblazed. The Core Technology Cluster for Innovative User Interfaces and Visualizations researches how semantics can improve the effectiveness of user-related tasks with information, e.g. searching and exploring information or imparting knowledge.

One of the main aspects as a Core Technology Cluster is the possibility of the visual adaptation to the requirements of the different Use Cases in the research program. Therefore the visualizations are subdivided into three layers, which allow an adaptation to different aspects of the requirements. From a visualization point of view, the layers layout and presentation are essential. The layout of semantics defines the visual representation of structures and placement at the user interface. Presentation covers icons, colors, or geometric shape and herewith, provides images best suited for the man-machine interaction interface.

In the technology session "Visualizing Semantics", a semantics visualization framework and its components will be presented to show the use of semantics in user-centered applications. The focus is on layout/placement algorithms and presentation techniques and their adaptation to different Use Cases. This session presents graphically implicit and explicit semantic information as will be modeled in the followed session "Ontology Design & Ontology Management".

ST3: Ontology Design & Ontology Management; Catherina Burghart, Forschungszentrum Informatik FZI
Stephan Grimm, Forschungszentrum Informatik FZI
Michael Sintek, DFKI

Ontologies constitute a technology for the explicit representation and handling of domain knowledge models in information systems. The W3C provides standardised languages for expressing and processing ontologies in the context of the Semantic Web, namely RDF(S) and OWL. While RDF(S) is often associated with the modelling of light-weight taxonomical hierarchies over classes and properties, OWL covers semantically richer language features for expressing logic-based statements, making each language suitable for a specific application context. Related to conceptual modelling in the area of object-oriented software development, the creation and maintenance of ontologies is a sophisticated task that requires effort and expertise in knowledge engineering. In this workshop, we will give a comprehensive overview on the W3C ontology languages RDF(S) and OWL and elaborate on their differences. We will be concerned with issues of ontology design and the how-to of modelling in RDF(S) and OWL. Moreover, we will cover the usage of standard tools for the handling of ontologies.

ST3+ST5: (Joint Session)
Applications and Benefits of Ontologies; Michael Grueninger, University of Toronto
Paul Buitelaar, DERI, National University of Ireland

In the final ontology session of the day, invited experts Prof. Michael Grueninger and Dr. Paul Buitelaar will present successful and promising approaches demonstrating benefits of using ontologies. The session will conclude with an open discussion on the future use and research directions of ontologies.

ST3: Service Description Languages; R. Kilian-Kehr, SAP AG
H. Weber, Fraunhofer ISST
Ch. Petrie, Stanford University

The "Internet of Services" aims at offering capabilities for the provisioning, discovery, individual use, and composition of services. In the longest range vision, discovery and composition can be performed on-demand and automatically in order to achieve current goals. Such automation may be viewed as a "World Wide Wizard" that would enable individuals to use services to "program the world" analogous to the use of installation wizards for PC software. Even without such automation, services should be described in such a way that individuals can easily find and consume them. Even better would be a way for people to describe and offer their own services easily.

Even for people to work directly with services, they will need computer tools for discovery, analysis, and use of services. So whether with the extreme vision of the WWWizard or human use, services must be described formally to enable their computer-supported definition, use, and composition. Current web service description languages, such as WSDL, are defined formally for their syntax only and for their semantics informally or semi-formally.

This was acceptable for the use of web services as a software engineering tactic but unacceptable for automatic use and composition of services. We also need more general notions of services to be formally described. Another view is that services may be sufficiently described on a programming level but need also to be described on an architecture level and most likely also on a user level. The language concepts needed on these different levels will differ but will still need to be mapped onto each other.

In this workshop we would like to discuss different service description concepts, their virtues and limitations, with the aim to get to a consensus that may be a basis for a standardization action.

This workshop will take into account existing semantic description standardization efforts, such as OWL-S, and WSMO, and will especially explore additional types of descriptions needed for various technologies needed for composition. The workshop will also include issues of the necessity, potential, and limitations of AI planning for automatic composition, and the semantic annotations required.

The workshop will also consider associated semantic annotations for so-called "non-functional" properties of services such as trust, quality of service, and service-level agreements. The workshop will also consider issues such as the degree to which service discovery and composition can be automated on the open Internet and the degree to which it may be restricted to "industrial service parks" where the semantics are agreed upon only within restricted communities.

Other highly-relevant issues of service descriptions to be treated include:

• business level aspects such as pricing model(s), legal aspects, and organizational aspects of services
• service bundling (and its relationship to business model modeling)- service variability modeling essentially allowing for service configuration and customization and simplified service consumption
• service extensibility and grey box approaches to service models
• aspects supporting service monitoring
• related protocol aspects of services (maybe based on pre-postconditions)
• relationship to domain-specific ontologies, e.g. industry-specifics

Finally, not only do we need semantic service descriptions, but we need descriptions that are easily used by humans. Service descriptions should as simply as possible describe the common elements of the world that services need to know and will affect, such as date, time, weight, and geography. Proposals for a business language of service descriptions are invited.

ST5: Semantic Image Annotation; Martin Huber, Siemens
Manuel Möller, DFKI
Pinar Wennerberg, Siemens
Daniel Sonntag, DFKI
Paul Buitelaar, DERI, National University of Ireland

Semantic annotation of medical images is a core element of the THESEUS use case Medico. In several talks, we will present manual and automatic ways of generating semantic image annotations both from the images themselves and from corresponding radiology reports, and how semantic image annotations can be modeled and used to improve image retrieval. Finally, comparisons to other imaging domains will be discussed.

APPLICATIONS

Multimedia Search Technologies

Libraries and media archives, such as those maintained by broadcasting stations, contain vast amounts of multimedia content. Typically this content is difficult to search efficiently, so that research activities and media access often require considerable effort. This is due to descriptive metadata, which is essential for identifying and searching for a medium, often being incomplete and of low quality.

In order to improve the situation, the development of media processing and analysis technologies is necessary. Currently, media archives still use mostly manual approaches to pre-processing (such as quality control) and content analysis, but those are prohibitively expensive for large archives. As a consequence, automated technologies for the restoration and analysis of multimedia content are essential to enable better access to archive content.

In addition, the inclusion of semantic technologies has been shown to improve searches considerably. With their help it is possible to not only find content, but also to get additional information about its context and its relationships with other sources of information.

This session will present several key approaches and technological advances in the area of multimedia search technologies, in particular restoration, content analysis and semantic searches.

With contributions from:

• Dr. Jan Hannemann, Deutsche Nationalbibliothek
• Sebastién Weitbruch, Deutsche Thomson
• Henrik Kinnemann, Siemens AG
• Patrick Ndjiki-Nya, Fraunhofer HHI
• Peter Altendorf, Institut für Rundfunktechnik
• Claudia Nickel, Fraunhofer IGD
• Joachim Köhler, Fraunhofer IAIS

Jan Hannemann, Deutsche Nationalbibliothek:
The CONTENTUS-Project: towards semantic searches for multimedia archives

Making multimedia archives searchable is a complex process. Typically digitization, restoration, content analysis and preparation for an efficient search access are among the required steps each medium has to pass through. The research project CONTENTUS aims to develop new concepts and technologies to facilitate the individual processing steps involved. The goal is an efficient workflow that starts with the original multimedia content and ends with its semantic linking and access in the context of a multimedia search. This talk will present the CONTENTUS project and illustrate how multimedia archives can benefit from these new information processing technologies.

Sebastién Weitbruch, Deutsche Thomson:
Efficient preparation of A/V content for the semantic multimedia search

Our Cultural Heritage includes archive holdings in the form of literature, music recordings, film and video. This cultural heritage is in danger, because it is threatened by decay. Moreover, with the advent of the internet age, these archive holdings move into the focus even further. In a world, where all information of mankind lies at the fingertips, it becomes more and more difficult to explain why there are cultural assets are decaying in basements, unexplored an inaccessible.

The German government has decided to change this situation, to develop technologies for making these archive holdings accessible, by initiating the THESEUS project. Archive holdings are digitized, digitally restored, descriptive metadata is extracted, and semantic metadata is generated using ontologies, ready for access via semantic search.

In our presentation we will discuss the transformation from analog content to high quality digital data and metadata for the semantic multimedia search, focusing on broadcast audio-visual archives. We will discuss the condition of these stocks at German Public Broadcasters, and what needs to be done to turn them into a searchable knowledge base. We will show what needs to be done to restore the picture content, and what metadata will be extracted and how, in order to turn these holdings into a repository ready for the semantic search. We will also discuss the workflow needed for content and metadata preparation. Furthermore, we will discuss how a semantic search can be realized.

Henrik Kinnemann, Siemens AG
Usage of Knowledge-Based OCR Technology for the Preservation of Cultural Heritage

Optical Character Recognition in combination with domain specific knowledge is crucial for understanding text in its context. Hence, it naturally plays a vital role in the European Digital Library Programme for the preservation of historical print media and video assets.

The presentation will focus on recent research on base technologies for

• retroconversion of born-digital text documents such as library file cards
• recognition of hand writing using comprehensive language models
• enrichment of regained document structure using automatic structure analysis in combination with visual analytics
• finding and recognition of text fragments such as subtitle and logos in video assets for annotation purposes and
• optimization of system parameters of restauration and annotation systems for large video image archives.

The talk will also try to stimulate discussions about possible future business models and the ways of dissemination of these combined technologies for digitization and semantic search.

Patrick Ndjiki-Nya, Fraunhofer HHI
Automatic Quality ensurance of Audiovisual Archive Contents for the Optimization of Semantic Search Engines

Within the THESEUS program, technologies for quality analysis of images and videos are being developed. These technologies are capable to detect defects such as compression artefacts in images and videos. Furthermore, quality features such as sharpness and contrast of images or videos can be assessed. These hints are used for restoration or for retrieval of high-quality content in multimedia databases.

Peter Altendorf, Institut für Rundfunktechnik
Metadata for Audiovisual Media - Possibilities and User Needs

Over the years, the manual annotation of audiovisual content at broadcasters has been optimised to a maximum cost-benefit ratio using standards and bodies of rules. Lately, the well-organized workflows have been further improved using file-based video production to allow a more efficient retrieval of archived content. Now, recent developments in data mining and content analysis could help to support the process of manual annotation, leading to a faster generation and more detailed metadata for the retrieval and reuse of content. But do these tools meet the user requirements? Has the "semantic gap" been bridged, yet? This presentation gives an overview of manually and automatically generated metadata in multimedia retrieval in the scope of CONTENTUS.

Claudia Nickel, Fraunhofer IGD
Perceptual Hashing for Identification of Multimedia Content

To identify multimedia content in large databases Perceptual Hashes can be used. Perceptual Hashing algorithms map the multimedia content to short digests which consider perceptual similarity. These allow the identification of the content even after small modifications such as compression or scaling are performed. In contrast to watermarking this is possible without a priori labeling of the content.

Dr. Joachim Köhler, Fraunhofer IAIS
Contentus: A Multimedia Semantic Search Engine for Cultural Content

The presentation shows the multimedia semantic search engine which is developed in the Theseus Use Case Contentus. The content from different media types (e.g. books, audio, images) is automatically indexed and the generated metadata is applied for search and browsing tasks. For the automatic indexing scalable methods and tools for document (OCR) and audio processing (audiomining) are shown in an integrated processing workflow. The output of the metadata generation process is stored in an intelligent repository system. Based on intelligent textmining technology the extraction of important information (named entity recognition) and text classification and clustering mechanism are used to support semantic search capabilities. The entities are disambiguated and linked external ontologies like WikiPedia to enable an intelligent browsing and the generation of useful additional search criteria. The presentation includes several demonstrations with cultural content from publishers and broadcasters. Users are enabled to navigate in large and inhomogeneous collections of multimedia content.

Kickoff Talks

ENTERPRISE COLLABORATION

INNORAISE - Know who knows;
Dr. Raphael Volz Managing Director, Innoraise

cm|navigator - fast and agile integration of Social Software into business processes;
Stefan Hauptmann cm|d - corporate mind development

Web 2.0 - collaboration tools like Wikis, (Micro-)Blogs, Forums, etc. are going to become an important part of the corporate IT-infrastructure. But establishing them and running them successfully very often fails. Wikis become orphan, blogs effect the opposite of their intention, etc. This is not an issue of technology but of corporate culture.

By applying our management and analysis framework cm|navigator we are able to identify the technical and cultural environment for the sake of good collaboration. By steadily employing workforce analysis and mirroring the outcomes to the involved collaborators we are able to set up and accompany paths of steadily growing and successfull collaboration.

MOBILE COMMUNICATION SERVICES

Enhanced ontology development toolkits for mobile communication services
Dr. Ilka Miloucheva,EU IST NETQOS Project

This talk is aimed to discuss requirements and design issues for enhanced ontology and knowledge engineering toolkit for mobile communication environment considering today and Future Internet infrastructures.
The motivation is based on experiences with ontology development for NETQOS autonomic QoS policy management system operating in heterogeneous Internet networks (EU IST project INTERMON).
The NETQOS modular ontology design and ontology classes, as well as their implementation using Protege-OWL are briefly overviewed.
Based on the evaluation of the NETQOS experience and the current state-of-the-art of ontology engineering technologies, challenges for enhanced ontology engineering toolkit for mobile communication services are derived.
The integrated toolkit is aimed to provide intelligent ontology engineering infrastructure for mobile communication services considering autonomic system approaches, domain specific and foundational ontologies, enhanced object oriented ontology presentation, as well as advanced knowledge infrastructure for mobile communication.

INTELLIGENT INFORMATION ACCESS

Knowledge Based Search – Towards Answering Questions (Download);
Dr. Michael R. Alvers CEO Transinsight GmbH

Does the world really need another search engine?" Our answer is YES. Why? Because today's search technology is in it's very early stadium. Today' engines allow full text search for words but they all fail miserably when it comes to more complex topics like biomedical research. The key for enabling search machines to communicate with the searchers is the intelligent usage of knowledge.
Transinsight developed GoPubMed. It is a Biomedical search engine which uses background knowledge in form of Ontologies for searches and leads faster to much better search results. The estimated time savings for knowledge searches are at least 50%. Time savings help scientists to concentrate on research rather than searching, is the first objective our company.

Semantic Technologies in a Corporate Research Portal (Download)
Philip Dudchuk, AviComp Services / Ontos, Oleg Ena, AviComp Services

Analysts of large companies usually face a whole range of tasks involving considerable volumes of routine 'manual' work. This leads to non-optimal time consumption, quality decrease, and ultimately loss of revenue. In this talk we will discuss a semantic solution which has been developed on the basis of a semantic API and implemented for one of our customers to solve these problems.
The use case under consideration is oriented at meeting the pains of several departments: research expertise, market and investment analytics, public relations, etc. The core of this system is a natural language processing engine which extracts semantically salient entities (people, companies, research terms, stock market information) and relations between them (invest, own, make a statement about, etc.). Merging same and similar entities allows to provide the following features:

• Semantic search across indexed content (documents from the Web and corporate libraries)
• Navigation through entities and relations all over the indexed content
• Creating and exploring profiles for organizations, people (experts), competences, research teams identified in the content
• Visualization of the network of semantic relations for a given entity and navigation through the associated documents
• Collaborative tagging and commenting on documents
• Subscription for semantic digests via e-mail or RSS feed (on the basis of a user's triple query, e.g. 'documents with all negative statements about the person X')

The system is implemented as a corporate portal built as a set of Web services meeting W3C semantic standards. This allows for easy integration with other customer applications.

Information Access based on open Standards – SMILA goes Eccenca (Download)
Hans-Christian Brockmann, Brox IT Solutions

The enterprise of things and the enterprise of services will eventually converge with enterprise processes to form the semantic enterprise. This vision requires the ubiquitous availability of semantically enriched information throughout the entire corporate value chain or the entire life cycle of a product. There might be many ways to achieve this goal. The SMILA consortium is convinced, that there is no better way to achieve this goal, than to provide an open standards based data integration platform that is shared by all vertical solutions.
Brox and its SMILA collaborators as well as the Theseus-Consortium are convinced, that this vision can become a reality. This talk will provide information as to how your organization can benefit from or contribute to such a standard. It will also share with you the steps and measures that brox and its partners are taking beyond providing a mere peace of Open Source software, to make such a standard a commercial succesware.

USER ACTIVITY ANALYSIS

Reading Stream Analytics - Attention counts;
Dr. Philipp von Hilgers, Managing Director, Meetrics GmbH

The Internet offers information on diverse subjects. What is still lacking are technologies to minimize the effort of selecting relevant information. Up to now, relevant pages are those requested by many users or explicitly linked or tagged. A new approach like Reading Stream Analytics makes use of implicit indicators to determine the relevance of a page or document. Someone, who takes his time to read a piece of information, shows implicitly the importance of the information. Compared to the overall user base the number of users rating the importance of a page explicitly is usually not very high. In contrast, implicit indicators of relevance can be recored in much higher numbers and without the need to disturb users while using the Internet. The lecture will present the technique of Reading Stream Analytics and demonstrate some use cases.

MULTIMEDIA INDEXING & RETRIEVAL

Improved image search using visual image sorting and semi-automatic generation of semantic relations between images;
Prof. Dr. Kai Uwe Barthel CEO, pixolution GmbH / HTW Berlin – University of applied sciences

We propose a new image search system using keyword annotations and low-level visual meta-data to generate semantic inter-image relationships. Unlike other approaches the new system does not try to learn the degree of confidence between images and associated keywords. We rather propose to model the degree of similarity between images by building up a network of linked images. The weights of the inter-image links are learned from the users’ interaction with the system only. For each image search a set of candidate images is selected from a visually sorted arrangement of result images. This candidate set is used to refine the search result by retrieving visually and semantically similar images. Semantic inter-image relation-ships of images can be modeled by collecting the candidate sets from many searches. Our system improves image search significantly.

From Retail to Photobooks: Semantic Retrieval in Large Consumer Photo Collections (Download);
Jochen Meyer, Director R&D Division Health, OFFIS GmbH

Quality Assessment of Semantic Image Annotation Algorithms; Peter Dunker, Head of Semantic Audio-Visual Systems, Fraunhofer IDMT

In this talk we present an image annotation challenge organized by the Fraunhofer IDMT (imageclef.org/2009/PhotoAnnotation) in the Image Cross Language Evaluation Forum (ImageCLEF), which allows the usage of provided domain knowledge (ontology) in the annotation process to improve the annotation quality. The main question is: Can an ontology (hierarchy and relations) help in automated annotation? We present a new benchmarking measure, which considers the relations of the domain concepts and punishes ignored relations. Finally, we announce an International Workshop (imageclef.org/2009/preCLEF), on image processing and retrieval as well as evaluation challenges, where the current and future challenges can be discussed.

VITALAS – Video & Image Indexing and Retrieval in the Large Scale;
Raffi Enficiaud, INRIA

Searching for Situations in Video Sequences
Dr. Andreas Hutter, Program Manager Multimedia Technology, Siemens AG

As a part of CTC WP2 an "intelligent" video retrieval system is being developed, which is able to solve semantic queries by jointly exploiting (inter-) relations between objects / events across multiple images, the situational context, and the application context. The innovation of the approach results from the combination of available video recognition components with structural interference, uncertainty-aware reasoning and context-based object content similarity queries. The challenges encountered and an architectural approach will be presented. The current implementation status will be discussed along with use cases for medium complexity queries operating on video surveillance sequences.

Exhibition

Building applications and infrastructures for a semantically enriched internet of services still needs research and development. Hence, there are lots of useable and lots of auspicious applications demonstrating, that the vision of the internet of services more and more becomes reality.

Some of those applications and demonstrators will be presented in our exhibition during the 2nd day of the symposium.

Exhibitors

• Avicomp Services & Ontos
• BOC Asset Management GmbH
• Brox IT-Solutions GmbH
• Deutsche Thomson
• DFKI - German Research Center for Artificial Intelligence
• empolis
• Fraunhofer-Institute for Industrial Engineering IAO
• Fraunhofer-Institute for Telecommunications, Heinrich-Hertz-Institute
• Innoraise
• Meetrics GmbH
• pixolution GmbH
• SAP AG
• Siemens AG
• TU Munich
• Vitalas Consortium