Science Award for Austrian Intelligent Knowledge Management application for e-Government DYONIPOS
07.11.2008
In September 2008, the Austrian research project 'DYONIPOS Use case BMF' won in the category Science Award – Most Innovative E-Government Application which was part of the 8th e-Government competition under the patronage of the German Ministry of Internal Affairs. The use case was carried out at the Austrian Ministry of Finance (BMF) in collaboration with Know-Center Graz and m2n - consulting and development. It focused on the implementation of an intelligent knowledge management system that reduces the amount of effort needed for search and analysis within administrations. The competition within the Science Award was open to federal, state and local administration in Austria, Germany and Switzerland. Tassilo Pellegrini, Head of Transfer & Media at SWC, spoke with Michael Granitzer (Know-Center), Doris Ipsmiller (m2n) and Josef Makolm (BMF), who were involved in all stages of the use case.
In your respective role as researcher, consultant and principal behind DYONIPOS: Can you give us a short insight into the application?
Michael Granitzer: DYONIPOS - which stands for DYnamic ONtology based Integrated Process OptimiSation - focuses on supporting knowledge intensive business processes in agile business environments. Basically, DYONIPOS automatically detects the current activity of a user and automatically delivers suitable information resources based on using the activity as context. We use an ontology based, semantic layer for providing a single-point-of-access to all information sources like e-mails, document repositories, yellow pages etc., thereby disambiguating existing information.
In addition, knowledge discovery technologies map unstructured artifacts into the ontology and expand the semantic layer by identifying relevant relationships. In contrast to other systems, DYONIPOS not only accesses server side resources, but also searches the client-side personal knowledge base of the particular user and offers both results.
This personal knowledge base is created automatically by the context observer module: if a user opens a particular document or web site of interest, DYONIPOS automatically integrates it into the personal knowledge base. Due to privacy reasons, such resources are added to the organizational knowledge base only after approval by the users. However, the sharing of resources is an important aspect increasing collaboration among knowledge workers.
To give a concrete example, consider that one of your colleagues informs you of a kick-off meeting for a particular project. By reading the e-mail, DYONIPOS detects that you may require additional information concerning this project and automatically provides you with a list of documents relating to the project, documents from similar projects in your personal knowledge base and a list of persons/organizational involved in the project or in topics covered by the project.
This is done automatically, so that there is no need for users to think about searching for in-depth information. We are very proud that the use case received the e-Government prize. The application of DYONIPOS in the Austrian Ministry of Finance shows how semantic technologies in general can be applied to such complex domains like e-Government.
Doris Ipsmiller: As one of the most important requirements stated by our pilot partner, the Austrian Ministry of Finance, knowledge management was not supposed to lead to additional work for the knowledge worker. Instead of forcing him to specify and provide further details about his knowledge, it was designed to help the user perform his day to day work even better and even more efficiently.
That’s why DYONIPOS interprets the current task of the user, deduces the information need and delivers important knowledge resources like documents, fellow employees concerned with the topic, related organisations and departments just in time. The user does not have to search explicitly, he does not have to know in which repository some piece of information may be found and he does not have to know how to formulate a query against the repository.
This is done by DYONIPOS automatically and “on the fly”. DYONIPOS adapts to the users’ information needs and learns how to fulfil them. The pilot implementation at the ministry has been turned into a productively running system, now being rolled out in the IT department within the ministry as a first step.
So DYONIPOS is far more than a functional, scientific prototype, but a fully applicable system dealing with many users and a huge amount of data. Semantic technologies and technologies of knowledge discovery have proven to be ready for widespread use.
Josef Makolm: It was the aim of the ministry, especially of the Directorate General for Information Technology (DG for IT), to set up a knowledge management system following a radically new, effective and economic approach. Classical knowledge management usually generates additional work. DYONIPOS is based on a proactive and knowledge discovery approach and activates the buried knowledge treasures hidden in the artefacts produced by the knowledge organization.
In addition, it was the aim of the ministry to set up a new innovation model for e-Government and to test a joint venture between science, economy and public administration, thus to initiate instant transfer of actual scientific results into practically usable applications and to shorten the time from scientific research to a usable e-Government application significantly.
Image below: the DYONIPOS project team, consisting of Michael Granitzer (Know-Center), Doris Ipsmiller (m2n), Keith Andrews (TU-Graz), Silke Weiß (BMF), Wolfgang Groiß (m2n), Josef Makolm (BMF).
DYONIPOS makes vital use of Semantic Web technologies like RDF and concept-related principles. How central have these ideas been to the realisation of the DYONIPOS application?
Michael Granitzer: Well, without using semantic technologies, DYONIPOS would not have been feasible. RDFS is used to harmonize the different repositories and integrate all kinds of different information sources for providing one single view on a company's repository. Also, user context and activities are represented in this semantic layer which in turn allows us to mine user activities for particular process patterns and to give means for optimizing workflows between knowledge workers.
Also, knowledge discovery and text mining technologies give us the possibility to enrich this semantic layer. Persons, organizations and locations as well as important concepts are extracted and added to the RDFS layer which again allows finding interesting relationships between persons and topics they work on.
So for example my person would be associated with the project DYONIPOS automatically, which is valuable information for other users who know nothing about me. Those relations are used in the background for improving information access, but are also directly displayed for browsing in the graphical user interface. Different kinds of visualization techniques help users to analyze different aspects of the knowledge base behind.
For retrieval, we also combine the power of structured SPARQL queries with an unstructured full text search. Unstructured searches are automatically expanded to deliver more relevant results. Also facets from the domain ontology are represented to summarize search results and can be subsequently used for filtering. Searching for user activities is one example for the power of semantic technologies. Since we represent user activities in RDF, we can easily apply all of the DYONIPOS retrieval mechanisms to search and analyze user activities. This is particularly of interest for roles in an organization interested in optimizing business processes or in gathering best practices.
Doris Ipsmiller: In short, the whole application makes use of RDF and OWL in order to represent any kind of structured data. Meta data from the native repositories is semantically represented and its concepts semantically harmonized, unstructured data like files and other artefacts are being semantically enriched and every kind of information derived from the use of DYONIPOS by the knowledge worker is represented in RDF/OWL.
Moreover, the internal program flows are modelled and expressed using semantic languages. Without semantic technologies, DYONIPOS could not have been developed and set into action.
Josef Makolm: One of the duties of the Directorate General for IT is to monitor scientific progress and to evaluate possible transformations into practical e-Government applications. As a result of this evaluation, we have decided to use semantic technologies for a completely new approach of knowledge management. Together with our partners we set up DYONIPOS as an utmost innovative e-Government application.
One of DYONIPOS' innovative ideas is to use graph-based principles, not just to model content but also for the application itself. How did the idea come around and does it scale?
Michael Granitzer: That's true. Above, I talked only about harmonizing data, but also workflows and processes are defined in RDFS. In particular, DYONIPOS is build on m2n Intelligence Management, short m2n IM, a framework developed by our partner company m2n. The IM framework is an ontology-based application development framework which allows building applications by modelling them in RDFS. For example workflows can be modelled using RDFS and executed via the m2n workflow and execution engine. There is no need to pre-compile workflows since they are interpreted on the fly and thus can be changed easily.
However, performance critical components like clustering or visualization are native components written in Java, but they are combined via workflows modelled in RDF. This synergy makes it extremely powerful: m2n's inference engine exploits the model semantics yielding to a very high flexibility. There is no need to have a programmer implementing changes to an application. Power users can learn how to change workflows easily by using m2n's graphical modelling environment.
Especially this flexibility in adapting to changing user requirements surprised the project leader of the ministries use case project, Josef Makolm, during the development of DYONIPOS. We did a lot of use case testing with employees in the Ministry of Finance and as usual each use case revealed a bunch of changes. Those changes could be easily integrated by adapting the workflows nearly on-the-fly, reducing the turnaround time for each use case test dramatically.
Doris Ipsmiller: Not only the program flows, methods, activities and mappings are represented in RDF but the entire data defining the application. The user interface for example is modelled using semantic technologies as well as access management and the data network for combining data semantically out of various repositories. In the project, we took advantage of the resulting flexibility in order to adapt the system quickly to the key user wishes. m2n IM allows modelling the whole application “on the fly”. The changes made are interpreted without restart of the system and without compilation of code.
DYONIPOS has to be able to constantly run parallel to the other user applications without slowing down the user system, has to deal with many users working upon one integrated knowledge base and has to handle large amount of data. So performance and scalability were very important aspects, but together we managed to tackle these challenges on diverse application layers. Otherwise it would have simply not been possible to provide our customers with responsive, scalable applications.
Josef Makolm: The use case project in the ministry turned out as chance to scale the used technologies and to achieve significant performance enhancements. The m2n Intelligence Management enabled us to model DYONIPOS in a very agile way; key users could try out diverse variations of the DYONIPOS system without any delay.
What are future plans for DYONIPOS? Are there other application areas outside the e-Government domain, where this method of software development can or will be applied?
Josef Makolm: Because of its modular design, DYONIPOS is set up as a kit of building blocks. This gives us the possibility to adapt DYONIPOS to various organizations as well as to integrate it into other systems.
Michael Granitzer: Especially the flexibility of DYONIPOS does not restrict its application to a particular domain. Using ontologies in the core allows us to easily adapt to different domains like for example competitive intelligence and of course we plan to do so. Of course there is always the need in fine tuning algorithms, especially if it comes to retrieval or knowledge discovery. But due to the easy adaptation we can focus nearly purely on the tuning aspect.
From a technical side, we will focus on extending our knowledge discovery algorithms for increasing accuracy further as well as in the extension of tools for doing development in RDF. Sophisticated, easy to use interfaces are needed to empower the user in changing workflows or developing new ones for her particular purpose.
While there is enough room of improvement, our underlying technology serves as a solid base to progress fast and so I am looking forward to new, fascinating application areas.
Doris Ipsmiller: m2n Intelligence Management as well as the Know Miner were brought together within various projects and are both designed as frameworks in order to build specific applications. So flexibility and adaptability to fulfil various requirements are very important aspects.
DYONIPOS is one specific implementation (with additional techniques like the context observer and just in time delivery). The underlying technology is being used in various other projects not only in the governmental but for example also in the industry sector.
About our interview partners
Dr. Michael Granitzer is Assistant Professor at the Knowledge Management Institute, Faculty of Computer Science at Graz University of Technology. He has been leading the division of Knowledge Relationship Discovery (former Knowledge Discovery) at the Know-Center Graz since 2004, where he is responsible for the acquisition and management of application oriented research projects as well as for the scientific strategy of the division. His scientific competencies are in the area of Information Retrieval, Machine Learning, Information Visualisation, Knowledge Discovery, Semantic Technologies and Multimedia Standards and Metadata. From 2001 to 2004, he was project leader in projects at the Know-Center covering topics like automatic text classification, text retrieval in millions of documents and visualisation of several million documents. He has led several work packages in the FIT-IT Project MISTRAL and managed the FIT-IT project DYONIPOS. Michael Granitzer has studied Telematics at Graz University of Technology with a special focus on Computational Intelligence and holds a PhD degree in technical science. He has been member in several program committees, organizer of several workshops and has published around 70 mostly peer-reviewed publications including journal publications, book chapters and books in the above mentioned fields.
Doris Ipsmiller is CEO of the innovation and development company m2n – consulting and development gmbh. She founded m2n in 1999 while being staff member of the Johannes Kepler University in Linz. Apart from widespread project experience, primarily in the public and industry sector, Doris Ipsmiller has lectured on the topics of knowledge management and knowledge organisation in diverse academic institutions like the Johannes Kepler University Linz, the Danube University Krems, Hagenberg University of Applied sciences and the University of Applied Sciences, Berlin. She has held speeches at various events and conferences like I-KNOW, IRIS, International Legal Informatics Symposium or the Symposium “Medienzeitreise” (Media Time Travel) in Berlin and has published numerous titles on topics concerning agile business process development, applied ontology management, ontology based application development and rule based engineering. She is currently leader auf numerous projects like the DYONIPOS Pilot project at the Ministry of Finance.
Josef Makolm is head of IT-Audit in the Directorate General for Information Technology at the Austrian Federal Ministry of Finance. He has over 30 years of experience in research, consulting and managing projects. His main activities and responsibilities are in e-Government, e-Taxation, e-Participation, e-Procurement, Knowledge Management, Interoperability and Multiple Use. He has published articles and books on these topics. He is member of the board of the Austrian Computer Society (OCG), co-leader of the Forum e-Government, head of workgroups in the Forum e-Government and the Austrian BLSG-Cooperation (cooperation including the federal state, the level of the provincial states as well as cities and municipalities) and lecturer at Danube University Krems. At present, he is project manager for the distinguished use case project DYONIPOS and the program leader of the Austrian part of the EU project PEPPOL (Pan European Public Procurement Online), where he is responsible for the Work package 2 "Virtual Company Dossier" - a project for borderless collection of business certificates and attestations.
Relevant Links
m2n consultung and development gmbh