Seminars - Invited Talks

Natural Language Understanding, Semantic Web, and Knowledge Engine – implications, challenges and opportunities

April 4th 2011

There is a rapid advance in areas of
natural language understanding, semantic web, and knowledge engine, as evidenced
by the emergence of deep understanding systems, semantic web based digital
libraries, expert finding systems, etc. Today, a well crafted deep understanding
system, e.g. IBM Watson Deep Q/A system, can outperform human in some quite
sophisticated natural language Question/Answer tasks such as Jeopardy, a
milestone in human-machine interaction. These advances have a profound impact to
human-machine interaction systems, and the infrastructure of Web where
information and contents are shared.

However, as with other emerging
technology trends, these emerging applications based on natural language
understanding, semantic web, and knowledge engine create new challenges and
opportunities. In this talk, we will overview the technology trend and new
development in these areas with a focus on communication, collaboration, and the
infrastructure of Web.

Answer Set Programming, the Solving Paradigm for Knowledge Representation and Reasoning

February 24th 2011

Answer Set Programming (ASP) is a declarative problem solving approach, combining a rich yet simple modeling language with high-performance solving capacities.  ASP allows for solving all search problems in NP (and NP^NP) in a uniform way (being more compact than SAT).  Applications of ASP include automatic synthesis of multiprocessor systems, decision support systems for NASA shuttle controllers, reasoning tools in systems biology, and many more.  The versatility of ASP is also reflected by the ASP solver clasp, developed at the University of Potsdam, and winning first places at ASP'09, PB'09, and SAT'09.

How to Value Software in a Business, and Where might the Value Go?

October 27th 2010

This presentation is intended to provide an understanding how software products are valued in the marketplace. This topic is rarely addressed in Computer Science education, to that few computer professionals are aware of the economic value of their products. The assessment of the value of their work has been left to business experts, economists, lawyers, and promoters. The lack of understanding hinders rational decision-making for tradeoffs in software design and implementation, market timing, and choosing business models. Having simple, quantitative models enables substantiating and revision of decisions made when planning software products.

The presentation will introduce some relevant business terms and concepts so that software engineering practices can be related to financial outcomes. In particular, estimation of software life, development lag, and generation and maintenance of intellectual property will be covered. To quantitatively compare alternatives some simple spreadsheet results will be discussed. The understanding gained has applicability to many areas of high-tech product design, resource acquisition, production, marketing, selection of business structures, outsourcing, and even the impact of taxation policies. It will also elucidate where current computer science education falls short.

WebId and the Social Web

September 8th 2010

Social Networks are now being used by billions of people and not a day goes by without them hitting front page news - with debates reaching up to the US Congress [1], as when Facebook recently changed it's privacy policy. But as everyone here in DERI knows these networks are data silos that make it impossible for cross site friendships to be established. What we can do with telephones, e-mail or the web, we still cannot do in the social networking space.

We know here that the global social networks can be created using linked data very easily. But those social networks are open for all to see. In order to allow the users to get the same feeling of intimacy as the social networks provide their users, we need to solve the global identity/authentication problem.

Henry Story will paint a quick picture of the seriousness of the problem, give a short background of the technologies that attempt to address it, and finally present WebID and show how it solves elegantly these problems. By using technology that has been available in the browser for the past 10 years and linking it to the semantic web, we get an extremely elegant solution to the problem of authentication, tying people in a viral way to the semantic web.


Computing for Human Experience: Semantics empowered Sensors, Services, and Social Computing on ubiquitous Web

September 7th 2010

Today, systems, devices, sensors, data and human participation enable something more than a “human instructs machine” paradigm. Traditionally, we had to artificially simplify the complexity and richness of the real world to constrained computer models and languages for more efficient computation. Now, mobile and fixed sensors as well as human-in-the-loop sensing, social computing and ubiquitous Web access work in concert to enrich interactions, information sharing and collective intelligence. Increasingly intelligent systems and participatory sensing capture observations that can be contextually integrated and enhanced to create awareness of events and situations— they not only deal with documents or entities but also support situational awareness by incorporating thematic (“what”), temporal (“when”), spatial (”where”), causal (“why”) and other relationships between objects and events. This positions us for what we call an era of “computing for human experience” (CHE) that supports a seamless interaction between the physical world and the cyber world, which encompass integrated capabilities in sensing, perceiving and recognizing the physical world (e.g., in extending sensory engagement with environments and narrowing the gaps between the real world and computing). It also uses “humans as sensors” of intensions and emotions, historical facts or background knowledge and community generated knowledge or collective intelligence. Semantic (Web) techniques and technologies (annotations, conceptual models/ontologies and reasoning) play a central role in important tasks such as building context, integrating online and offline interactions, and help enhance human experience with focus on natural activities while relegating explicit computing and communication activities to background.

In this talk we will give a brief background of Kno.e.sis and discuss instances of semantics-empowered services computing, social networking and sensor Web that point to early capabilities towards CHE. An article in IEEE Internet Computing provides further information:

Uncertainty and Vagueness in Semantic Web Languages

June 11th 2010

There is currently a strong interest in using and extending AI techniques, systems, and concepts to the World Wide Web. In particular, managing uncertainty and/or vagueness is starting to play an important role in Semantic Web research.

Our aim is at making attendees familiar with the concepts and techniques for representing and reasoning with uncertain and vague knowledge in current Semantic Web ontology and rule languages, such as RDFS, OWL, RIF, (and their combination), which should help the attendees to get insights on main features and practical and theoretical problems of the formalisms and tools proposed so far.

Semantic Approaches for Searching and Managing Medical Images and Text -- Research Project THESEUS-MEDICO

June 4th 2010

Medical images contain valuable information, which is not utilized, as this is not linked to the images in a machine-processable form.
Furthermore, the related textual data such as radiology reports or scientific publications are most often stored elsewhere, so that finding the coherent set of patient images and text becomes a significant challenge.

The German government funded research project THESEUS-Medico [1] addresses these problems by taking a semantic approach, which is based on annotating patient images and radiology reports using medical ontologies. The coherent set of data relevant to the user's query will then be retrieved on the basis of the annotations.

This talk will first introduce the MEDICO project. Then it will report on various knowledge engineering activities with particular focus on ontology modularization involving Radiology Lexicon [2] and Foundational Model of Anatomy [3] ontologies. Finally, it will outline natural language processing applied to radiology reports.


Social Networks and Interactive Management: interdependent problems and interdependent solutions

May 5th 2010

Many of the problems we face require knowledge and input from many individuals. The resolution of these problems often requires some modicum of higher-order relational or systems thinking that allow groups of individuals with a vested interest in common problems to see clearly the structure of their problems and design and plan action sequences that help them to resolve their problems in an efficient and effective manner. The systems science methods developed by John Warfield and his colleagues offer great power and potential in this context. Specifically, Interactive Management (IM) allows a group of individuals with a vested interest in solving a problem to design problematiques (i.e., graphical influence structures) that describe causal relationships between a large set of problems in a problem field. IM taps into and enhances our largely underdeveloped cognitive capacity for graphical, systems things. It enhances the collaborative power and action potential of groups who seek to work together toward the resolution of problems and the realization of possibilities. The IM method has been applied successfully in the resolution of manufacturing and service problems in American industry, conflict resolution between Greek and Turkish Cypriots, and the design of new infrastructure in China and Mexico (Broome, 2006; Warfield, 2006; Warfield & Cárdenas, 1994). With an understanding of their common goals, and with insight into the structure of their problems, a group using IM can design action structures that help them to solve their problems and achieve their goals. There is scope for IM to be integrated into a Web 2.0 interface, thus making IM more widely available to a variety of social networks and more functional for groups working at a distance. There is also scope for IM to be integrated with web-based knowledge systems that facilitate groups making critical decisions during their design of influence structures. I advocate a Pragmatic Web of Systems Science Action, a project envisaged as a collaboration between the Centre for Innovation and Structural Change (CISC), the Digital Enterprise Research Institute (DERI), the School of Psychology, and the School of Education. I believe that many of the “problems of development” we strive to solve collectively can be better solved if we work together using systems science tools.

Guidelines to Publishing your Research Output

March 1st 2010

One of the imperatives of scientific research and development is to exploit the final outcomes. This exploitation may take the form of marketable products, or in most circumstances, publication(s) in reputed conferences and journals. These publications, by virtue of going through peer review process, will reflect on your standing as a researcher. The talk will discuss various channels through which you can disseminate your research results. It will present a systematic way of writing scientific papers – the structure, the dos, the don’ts, etc. The material of the talk is primarily based on Hari’s experience as Editor of an International Journal.

The structure of the talk will be informal in nature and the attendees will be encouraged to ask questions. After the talk, if required, Hari will be available for one-to-one session or small group discussions.