PhD School 2008 in Innsbruck

The seminar is targeted at PhD students whose research interests cover the Semantic Web and Web Service areas. It will help the students concretize their research intentions and receive feedback on the soundness and completeness of their achieved work.

The seminar is open to everyone who's interested, it is not restricted to PhD students and their supervisors.



19. - 21.11.2008



This year the seminar will be a joint event between STI Innsbruck and the AIFB Karlsruhe. The seminar will take place in Innsbruck, 19.11.2008 - 21.11.2008 on the premises of STI Innsbruck.

Location: Room 3W03 of ICT building
2nd Floor, Technikerstraße 21a, 6020 Innsbruck. More information how to get there can be found here.

The seminar will contain presentations of 45 minutes (30 minutes talk + 15 minutes discussion) for all PhD students, independently of the phase they are currently in.


Important Dates

Deadline for research reports (only STI Innsbruck students) 03.11.2008

Seminar: 19. - 21.11.2008, Innsbruck


Detailed Schedule

12.15 - 14.00 Lunch  
14.00 - 14.15 Welcome
14.15 - 15.00 Tobias Bürger (STI Innsbruck)

A Conceptual Model for Intelligent Content for the Semantic Web
While the amount of multimedia content on the Web is continuously growing, reuse of multimedia content remains low and automated processing of content remains hard. The retrieval of multimedia content on the Web is a continuous challenge which is due to the lack of (formal) descriptions and generic multimedia analysis algorithms. Describing multimedia resources through metadata is thus often seen as the only viable way to enable efficient multimedia retrieval. We present an overview of an ongoing PhD thesis on a conceptual model for intelligent content. The model can be used for the description of multimedia content and associated information on the Semantic Web. The goal of the thesis is to improve multimedia content retrieval on the Web with a particular focus on retrieval of reusable multimedia content in professional domains. The working hypothesis of this thesis is that semantic descriptions exploited in semantic search and their combination with content-based retrieval provides better precision in professional multimedia retrieval.

15.00 - 15.45 Sebastian Blohm (AIFB)

Large Scale Information Extraction from the Web -- The online/offline trade-off
A large amount of information on the internet is made available in the form of text. It constitutes a pool of knowledge that is ever updated, valuable to understand trends and unparalleled in its coverage. Capturing information from text gives semantic access to the content produced by journalists, online communities and the marketing industry. When trying to access large portions of the Web, two types of techniques become relevant for efficient processing: - large scale data mining techniques must be used that allow to find scarce but valuable regularities in the corpus. - Information Retrieval techniques such as Web search indices and probabilistic corpus statistics need to be used to give access to interesting portions of text without having to process all textual content in a linear fashion.
At the same time, processing should capture form, structure and meta-information of texts. There is a trade-off between structure and text analysis that is done prior to information retrieval indexing (offline) and during information extraction or user interaction(online). I will present ongoing and completed work that takes various positions along this trade-off.

15.45 - 16.30 Mick Kerrigan (STI Innsbruck)

MEMOS: A Methodology for Modeling Services
Semantic Web Services are a new and emergent technology for improving the dynamism and automation in Service Oriented Architectures. Past experience has shown that such new technologies can be very disruptive when adopted in software engineering projects until methodological and tool support is provided. This presentation introduces the MEMOS methodology for modeling services semantically using the WSMO paradigm. It focuses on the key scenarios in which Semantic Web Services need to be created and the relevant software artifacts that need to be developed to enable Semantic Execution Environments.

16.30 - 17.00 Coffee break
17.00 - 17.45 Frank Dengler (AIFB)

Service value networks in enterprises
Today, many different knowledge structures are identified in enterprises, like operational structures, organizational structures, information structures, and social networks. But enterprises find it hard to transform much of their essential knowledge into transferable, easily accessible, and actionable knowledge assets, because such hidden knowledge is typically in these knowledge structures or held by individual knowledge workers. These knowledge workers connected to each other through enterprise knowledge structures can be seen as a service provider and a service requester. The main goal of this PhD thesis will be, how these interrelations between knowledge workers in enterprises can be mapped to service value networks and how these networks, and thus the information flow in enterprises, can be improved.

17.45 Closing
20.00 Dinner (Löwenhaus)
09.30 - 10.00 Coffee  
10.00 - 10.45 Srdjan Komazec (STI Innsbruck)

Towards Monitoring Semantic Execution Environments
An environment performing automation of the Web Service usage tasks, such as service discovery, selection and ranking, data and process mediation, composition, choreography and invocation, is refereed to as a Semantic Execution Environment (SEE). The tasks are typically delegated to a set of centrally orchestrated components. The complexities and relationships between the SEE middleware components introduce the necessity for a comprehensive observation activity in order to enable autonomic environment capable of a sustainable functionality. Furthermore, the uncertainties and failures caused by the presence of remotely accessible services (and the overall networked context) are additionally stressing the need for an end-to-end monitoring solution which can assist in various service related activities such as enforcement of Service Level Agreements, verification of functional compositions, measuring of non-functional properties and handling of faults. A possible solution envisions application of the event-based monitoring techniques integrated with the formal knowledge representation mechanisms in the context of Semantic Execution Environments. The talk will give a brief overview of the motivation and scientific background regarding the topic as well as the possible approach which will be taken to introduce reactive behavior in a SEE. The talk will also present research challenges, possible contributions, work and dissemination plans.

10.45 - 11.30 Uwe Keller (STI Innsbruck)
11.30 - 12.15 Tuvshintur Tserendorj (AIFB)

Approximate Reasoning Over Expressive Ontologies
With the increasing interest in expressive ontologies for the Semantic Web, it is critical to develop scalable and efficient ontology reasoning techniques that can properly cope with very high data volumes. For certain application domains, approximate reasoning solutions, which trade soundness or completeness for increased reasoning speed, will help to deal with the high computational complexities which state of the art ontology reasoning tools have to face.

12.15 - 14.00 Lunch
14.00 - 14.45 Eugenie Giesbrecht (AIFB)

Interactive Query Answering from Unstructured Resources using Semantic Metadata
Information retrieval is concerned with answering users' information needs as accurately as possible. The necessary information can be provided by ever-growing online resources. A traditional paradigm of keyword-based document retrieval is, however, unable to satisfy particular information needs of the user. Retrieval of documents contained in today's very large digital collections often results in a large amount of possible relevant results so that the user is overloaded with consulting lots of retrieved documents in order to find the answer to his or her information need. So, we need to redefine the retrieval paradigm, which can be explored in the following two directions.
On the one side, information extraction technology offers the extra "knowledge" to include in document and query representations. On the other side, there is always a unique specific user behind the query, and if the answer to the query is not straightforward enough, it may be necessary to negotiate with the user his actual information need. The latter is especially the case for non-factoid questions, or in decision support applications relying on online information access.

14.45 - 15.30 Igor Popov (STI Innsbruck)
15.30 - 16:15 Omair Shafiq (STI Innsbruck)

Semantic-enabled Coordination for Services across Web
It is becoming increasingly important that the Service Oriented Systems of the future should be able to flexibly adapt and deal with the dynamic changes that may occur in the distributed and large-scale environments like Web. Several existing solutions use the syntactic and limited information that is modeled and hence their ability to sustain in dynamically changing and open environment becomes limited. Therefore, middleware based solutions for SOA, i.e. Enterprise Service Bus solutions are limited to a closed environment and to a limited set of components and services. This thesis proposes a solution that allows the middleware based solutions for services (i.e. Enterprise Service Bus) to adapt to the dynamically changing environment, availability of services and components over the Web. The solution introduces semantic descriptions to the components in a Service Bus. It further proposes how this semantic-enabled information about the components can be shared in a Service Bus across Web using a global space-based communication approach. The semantic descriptions of components helps in precisely defining the descriptions of components and the semantics are modeled based on widely-accepted standards. Moreover, sharing this information across Web is carried out by a global semantic-space based communication paradigm, which is foreseen as a Web for machines. This thesis will present the model for semantically describing the components, algorithm for processing the semantic descriptions, a semantic execution engine for the service bus to process the semantic descriptions as well as interaction patterns for communication based on semantic-space based communication paradigm. The solution has been developed based on open Service Bus standards, and hence is implementation neutral. The thesis will further present that how this solution can be applied to currently available Middleware based solutions for Services, i.e. WSMX (Web Services Execution Environment). Evaluation plan also been built that shows how the proposed solution will help the Service Bus in handling newly added components, finding the right component with higher precision (out of multiple components available at a given time) and how the semantics will help in handling the complex execution workflows, as well as how the degree of decoupling between the components can be increased by allowing real-time binding of components during a workflow execution.

16:15 - 16:45 Coffee Break
16.45 - 17.30 Katharina Siorpaes (STI Innsbruck)

Games for semantic content creation
Despite great progress in technology, many tasks in semantic content creation still heavily depend on human intelligence and can not be completely automated. There is still an substantial amount of of human intervention required. However, we can also observe only very limited human involvement in many sematnic content creation tasks. In order to provide an incentive and motivate users to dedicate their time to semantic content creation, we propose to hide tasks behind online games. We have analysed tasks for semantic content creation for the degree of human intervention that is required to complete them. Then we select the tasks that are most suitable to be turned into games and describe possible games for semantic content creation. We describe the generic OntoGame platform as well as two concrete game scenarios.

17.30 Closing
20.00 Dinner and Party (Stadtcafe)
09.30 - 10.00 Coffee  
10.00 - 10.45 Reto Krummenacher (STI Innsbruck)

A Parametric Design Approach to Scalability Management of Semantic Space Middleware
In this work we discuss the requirements and challenges arising in the context of Web computing. As main contribution, we elaborate on scalability implications, and present a scalability-driven requirements analysis tool that supports developers, owners and administrators during the design phase. The objective of the tool is to ease the definition of realizable, scalable and dependable semantic space platforms that match the needs and expectations of large-scale distributed knowledge systems. Furthermore, it prevents developers from having to care about the technical details and the underlying complexities of the middleware system implementation. The tool is based on parametric design problem solving, for which the characteristics of the functional and non-functional properties, the trade-offs and scalability impacts are taken as input and reasoning constraints. The principle of parametric design is to assign distinct values to a pre-arranged set of components, called parameters. The parameters are specified in so-called templates and are instantiated by means of search algorithms (problem solving methods) in order to come up with a complete configuration.

10.45 - 11.30 Sebastian Speiser (AIFB)

An Economical Approach to Data Web Search
Current search methods aim mainly at maximizing the relevance of the results according to the user's query. It is common that users pose informal queries (e.g. keywords) which have to be translated to formal queries that the underlying data store can answer (e.g. SPARQL queries). Maximum relevance is often achieved by combining the results of multiple formal queries and create a mixed ranking. Another approach is to directly rank the formal queries and present the best ones to the user. In this way he can decide which is the correct interpretation of his information need. A further advantage is that only one query has to be executed which is inherently cheaper than evaluating several. Based on such a system by Tran et al. we develop an extension that takes the execution costs and expected relevance of formal queries into consideration. The assumption is that there is no interest in increasing relevance regardless of costs. An use case is keyword search on RDF data hosted in the cloud. The user has interest in getting high relevance but at the same time wants to keep costs low as he is directly charged for computing power and I/O usage.

11.30 - 12.15 Closing discussion
12.15 - 14.00 Lunch

Workshop Organization

Dieter Fensel (STI Innsbruck, Austria)
Elena Simperl (STI Innsbruck, Austria)

Webmaster: Mail:

Jump to top.