RUC Computer Science Research Seminars: Spring 2008
All seminars take place from 12.00 - 13.00 in
Room 42.1.37 except where noted specially. Note change to earlier time compared to previous semesters.
RUC Computer Science Research Seminars: Spring 2007
All seminars take place from 12.30 - 13.30 in
Room 42.1.37 except where noted specially.
RUC Computer Science Research Seminars: Autumn 2006
All seminars take place from 12.30 - 13.30 in
Room 42.1.37 except where noted specially.
RUC Computer Science Research Seminars: Spring 2006
All seminars take place from 12.30 - 13.30 in
Room 42.1.37 except where noted specially.
RUC Computer Science Research Seminars: Autumn 2005
All seminars take place from 12.30 - 13.30.
Location: Room 42.1.37
RUC Computer Science Research Seminars: Spring 2005
All seminars take place from 12.30 - 13.30.
Location: Room 42.1.37
RUC Computer Science Research Seminars: Autumn 2004
All seminars take place from 12.30 - 13.30.
Location: Room 42.1.37
RUC Computer Science Research Seminars: Spring 2004
All seminars take place from 12.30 - 13.30.
Location: Room 42.1.37
RUC Computer Science Research Seminars: Autumn 2003
All seminars take place from 12.30 - 13.30.
Note change of time. Location: Room 42.1.37
Keld Bødker and Kristian B. Bøving. WHERE IS THE INNOVATION? The Adoption of Virtual Workspaces
Abstract
This paper reports from a case study of the introduction of a web-based groupware application Lotus QuickPlace in a large European financial organization. Our study challenges the commonly held assumption in DOI research that all use is equal, implied when DOI is used to study the adoption of technologies. The underlying problem is that in order to understand the diffusion of groupware, we need to distinguish between two separate innovations. The first is directly related to the DOI "innovation-decision process" - centered on the technology as the innovation. The second innovation is more related to how the technology is put to use. This has consequences for devising strategies for the diffusion, and we suggest a strategy for diffusing virtual workspaces, which combines the DOI framework's one-way communication with the two-way communication of a participative approach.
Henning Christiansen: A new, constraint-based grammar formalism and analysis method for natural language processing
Abstract
We may discuss why on earth someone, and even a computer scientist,
dare to get the idea of suggesting new concepts and methods to a
well-established field that require an huge background knowledge about
language.
Well, some researchers start with a problem and try to solve it, others
study how other people solve their problems, ...
In the present research we claim to have found a solution, and the next
steps will be to find out which problems it may solve....
The seminar will be based on a presentation used at RANLP,
http://lml.bas.bg/ranlp2003/ in Bulgaria a few weeks ago.
It was entitled "A constraint-based bottom-up counterpart to DCG"
and the paper abstract goes as follows:
A new grammar formalism, CHR Grammars (CHRG), is proposed that provides a
constraint-solving approach to language analysis, built on top of the
programming language of Constraint Handling Rules in the same way as
Definite Clause Grammars (DCG) on Prolog. CHRG works bottom-up and adds
the following features when compared with DCG:
-
An inherent treatment of ambiguity without backtracking.
-
Robust parsing; do not give up in case of errors but return the recognized phrases.
-
A flexibility to produce and consume arbitrary hypotheses making it
straightforward to deal with abduction, integrity constraints, operators
à la assumption grammars, and to incorporate other constraint solvers.
-
Context-sensitive rules that apply for disambiguation, coordination in
natural language, and tagger-like rules.
John Gallagher: Soft Types
Abstract
Types in programming languages are generally
agreed to have a positive effect on program
quality, programmer productivity and efficiency
of compiled code. Usually, types are declared by the
programmer, in which case they are called prescriptive types.
If a program is not consistent with the prescriptive types
then the program is rejected.
Types can also be
inferred from a program (in either typed
or untyped languages) even when the programmer does not supply
any type declarations.
These are called descriptive or soft types.
Deriving soft types automatically is a static
analysis problem. Soft types tend to be more
application-specific, and can give us a lot of
useful information about the program's behaviour.
The talk will give examples of soft types in
various programming languages. Then there will be an outline
of procedures for deriving soft types. Finally some of the applications will be
sketched, such as
debugging, verification, and specialisation.
Keld Helsgaun: An effective heuristic algorithm for the traveling salesman problem
Note: The talk will be in Danish, with English slides
Abstract
The traveling salesman problem (TSP) may be stated as: A salesman is required to visit each of some given cities once and only once, starting from any city and returning to the original place of departure. What tour should he choose in order to minimize his total travel distance?
The importance of the TSP stems not from a massive need from salesmen wishing to minimize their travel distance. The importance comes from a wealth of other applications, many of which seemingly have nothing to do with traveling routes
Furthermore, the TSP is a typical problem of its genre: combinatorial optimization. This means that theoretical and practical insight achieved in the study of the TSP can often be useful in the solution of other problems in this area. In fact, much progress in combinatorial optimization can be traced back to research on the TSP.
In this talk I will describe a heuristic algorithm for solving the TSP. Computational experiments have shown that it is highly effective. Even though the algorithm is approximate, optimal solutions are produced with an impressively high frequency. It currently holds the record for a series of large-scale instances with unknown optima, among these a 1,904,711-city instance of locations throughout the world.
Davide Martinenghi: Simplification of Integrity Constraints for Data Integration
Abstract
To design and implement a data integration system means to provide a unified view (the global database) of a collection of data sources (the local databases). Integrity checking in the context of data integration is a delicate matter, because the consistency of the local sources is not a sufficient condition for the consistency of the global database.
Simplification techniques for integrity checking in traditional databases can be generalized to tackle the problem in an elegant and efficient way: the knowledge of local consistency is employed, perhaps together with other knowledge which is trusted at the global level, in order to obtain conditions that allow an efficient verification of global consistency. Furthermore, when a violation of integrity is detected, it is possible to repair it by maintaining a virtual database of exceptions at the global level.
In this talk we describe these techniques and show examples for both the local-as-view and global-as-view approaches to data integration.
Morten Rhiger: Types for Program Generation
Abstract
In this talk we address the challenges of type-safe multi-stage
languages.
Multi-stage languages support safe generation, manipulation, and
execution of program fragments. They are unavoidable in compiler
design, program specialization and optimization, and run-time code
generation. Despite recent effort, however, most existing multi-stage
languages do not fulfill their obligation to being both safe and
flexible.
Type-safe multi-stage languages must guarantee not only that
well-typed programs do not "go wrong" but also that any program
fragment generated by a well-typed program is again well-typed. The
reason is that such a program fragment may eventually be executed and
may therefore contribute to the overall execution of the program.
We present two multi-stage languages and discuss their type safety and
their flexibility:
The first is an example of an embedded language: It combines the
linguistic infrastructure of an existing meta language with a
simply-typed object language. We show how phantom types, a
programming idiom originating in functional programming, enables the
meta language to "host" the object language type system.
The second is a core multi-stage calculus designed from scratch. This
calculus resembles existing multi-stage calculi but it provides more
flexibility in the use of, e.g., side effects. The development of
this calculus is ongoing work. If time permits, we sketch how to
extend this calculus to multi-stage versions of Standard ML and Java.
Troels Andreassen, Rasmus Knappe and Henrik Bulskov: Ontology, Ontology-Based Retrieval, and Similarity
Abstract
The topic of this talk is information retrieval mediated by a generative ontology.
Documents are indexed on the basis of an ontology and queries are interpreted with respect to the same ontology. By these means documents and queries can be mapped to a unified form - descriptions formed by concepts of the ontology. We introduce firstly to the key problems in the approach. How is the ontology defined? How are documents indexed and queries interpreted? How are queries evaluated?
Secondly we go into more detail on the latter. The ontology is first of all to influence ranking of objects in answers to queries, as measured by each individual objects degree of similarity to the query. As in any IR approach the goal is to obtain the best possible ranking. Thus the goal in this context is to derive measures of similarity that respects and exploit the ontology.
Finally we present a prototype querying system that applies ontology-based similarity in a generalization of standard fuzzy information retrieval.
Morten Hertzum: Software Engineering: Managing Complexity or Muddling Through
Abstract
Software engineering is a complex undertaking that abounds in
collaboration and information seeking. Major constituents of this
complexity are the creative element entailed in articulating and
reaching closure on requirements, the progress imperative entailed in
making estimates and tracking status, and the collaboration challenge
entailed in learning within and across projects. These three
constituents of software engineering are discussed in relation to
software-project risks and illustrated with two empirical cases. The
talk concludes by outlining two areas for improving our understanding of
software engineering and devising better tools for software engineers:
collaborative information seeking and people finding.
Jesper Simonsen: Participative Design With Top Management: Anchoring Visions by the Problem Mapping Technique
Abstract
På vores sidste research seminar præsenterede Morten nogle "top-ti-lister" over problemerne ved IT projekter. Top-scoreren af problemer var ifølge Schmidt et al. (2001) "Lack of top-management commitment to the project".
Jeg vil præsentere en case fra et forskningsprojekt hvor netop dette problem blev løst ved hjælp den såkaldte kortlægningsteknik (diagnostic/virtual mapping).
Casen var et samarbejde mellem mig og IT-konsulenter fra WM-data ifm. det CIT-finansierede projekt "Arbejdsformer i datalogiske forundersøgelser" hvor MUST metoden blev afprøvet ('disseminated') til erfarne IT-praktikere.
Jeg vil præsentere casen og samarbejdet med WM-data, MUST metodens princip om forankring samt demonstrere brugen og effekten af kortlægningsteknikken gennem et detaljeret eksempel, hvor WM-datas IT-konsulenter anvendender teknikken med (top)ledelsesrepræsentanter hos en potentiel kunde - skattemyndighederne for et land i Europa.
Resultatet af forskningsprojektet har haft konkret indflydelse hos WM-data, for MUST metoden, i min nuværende forskning i HIT projektet samt i undervisningen på datalogi - ikke mindst i SUO kurset.
Abstract (fra artikel under publicering):
This article reports on a project introducing techniques from the MUST method for IT designers in a large international supplier of systems for tax assessment and auditing. The focus is on evaluating the fit between the supplier's system and the customer's requirements, particularly through meetings aimed at aligning top management with the supplier's analysis. The article describes the MUST method's anchoring principle and the technique of problem mapping supporting this principle. This participatory approach resulted in mutual learning processes with top management which is rarely reported on in the Participatory Design community. Top management participated by reviewing, challenging, and reformulating the IT designers' central suppositions, assumptions, and hypotheses related to the causal relation between identified problems and suggested solutions.
Jørgen Villadsen: Paraconsistent Logic in AI
Abstract
Classical logic predicts that everything (thus nothing useful at all)
follows from inconsistency.
A paraconsistent logic is a logic where an inconsistency does not lead to
such an explosion, and since in practice consistency is difficult to achieve
there are many potential applications of paraconsistent logics in Computer Science
and Artificial Intelligence (AI).
The talk gives an overview of the forthcoming presentations at the conferences
Artificial Intelligence and Symbolic Computation (AISC 2004)
and
Multiagent System Technologies (MATES 2004).
The first paper presents a concise description of a paraconsistent higher order logic with countably infinite indeterminacy.
The meaning of the logical operators is different from traditional many-valued logics as well as from logics based on bilattices.
Higher order logics are expressive and with several automated theorem provers available.
The second paper argues that paraconsistent logic is advantageous in order to deal with assertions made by intelligent agents.
Other propositional attitudes like knowledge and beliefs can in principle be treated along the same lines.
A first order variant of the paraconsistent higher order logic is translated into classical predicate logic.
Elvira Albert and Germán Puebla: Abstraction-Carrying Code
Abstract
Recent approaches to mobile code safety, like "proof-carrying code",
involve associating safety information to programs. The code supplier
provides a program and also includes with it a certificate (or
"proof") whose validity entails compliance with a predefined safety
policy. The intended benefit is that the program consumer can locally
validate the certificate w.r.t. the "untrusted" program by means of
a certificate checker---a process which should be much simpler,
efficient, and automatic than generating the original proof. This
talk presents a novel approach to mobile code safety which follows a
similar scheme, but which is based throughout on the use of abstract
interpretation techniques. In our framework the safety policy is
specified by using an expressive assertion language defined over
abstract domains. We identify a particular subset of the abstract
interpretation-based static analysis results which is especially
useful as a certificate. The validity of the certificate on the
consumer side is checked by a very simplified and efficiently
specialized abstract-interpreter. The resulting scheme has been
implemented and benchmarked within the CiaoPP system, which will be
demonstrated during the talk.
Keld Bødker and Morten Hertzum:
Muligheder og udfordringer i BSCW som støtteværktøj i projektarbejdet
Abstract
Med baggrund i vedlagte artikel
(bidrag til temanummer af UNEV)
og et mundtligt oplæg, vil vi invitere til en diskussion af BSCW
og/eller eCampus i vores undervisning.
Niels Jørgensen: Teaching Software Design: Top-Down versus Bottom-Up
Abstract
My presentation
will sketch ideas for future work on software design.
The general context is the broad question of how computer
science research contributes to software design and engineering?
More specifically, the (preliminary) research questions is:
what are the contributions of computer science to the models
that designers use to deal with design uncertainty?
The wing profile of an American WW2 bomber airplane
provides an illustrative design example from aviation engineering
(as analyzed by Vincenti); the wing design was successful,
but the causes of its success were not understood at first.
Now they are, and wing design in modern aviation engineering
is largely science-based. On the basis of this analogy,
a taxonomy of models used in software design is proposed.
Kim Henriksen: Analyzing PIC programs
Abstract
Software developed for pervasive systems has much stricter requirements to
resource consumption, correctness etc. than conventional software developed
for your PC. In most cases software
will be developed in assembler, since no high level languages can produce
code that meets the resource requirements. However, developing software in
assembler is typically more prone to programming errors, difficulties in
debugging and so on.
In this talk, I will present my approach for allowing program analysis
tools and techniques, that have been developed with higher level languages
in mind, to be used on programs developed in a low level language.
Torben Bräuner: Modal logik: Fra oldtiden til det 21. århundrede
Abstract
Modal-logiske argumenter kan via middelalderens logikere spores helt tilbage
til oldtiden, bl.a. til Aristoteles' skrifter. Det var dog først i
1950'erne og 1960'erne, modal logikken blev fuldt ud formaliseret,
herunder blev givet den nu alment accepterede Kripke-semantik.
Målet var da at kunne formalisere filosofiske argumenter
som involverede f.eks. begreberne tid, nødvendighed og viden.
I løbet af 1980'erne og 1990'erne fik modal logikken et bredere sigte,
nemlig at være et generelt sprog til at ræsonnere om relationelle
strukturer - et alternativ til standard første-ordens og
højere-ordens logik.
Datalogien leverer mange eksempler på sådanne relationelle strukturer,
f.eks. træ-strukturer, men også matematikkens naturlige og reelle tal
er to prominente eksempler. Standard modal logik har svagere udtrykskraft end
første-ordens logik, men er i modsætning til første-ordens
logik afgørlig. I den nyeste forskning spiller modal logik - og udvidelser
heraf - faktisk en betydelig rolle i udforskningen af grænseområdet
mellem det afgørlige og det uafgørlige. Såkaldt hybrid logik
er et eksempel på en sådan udvidet logik fra foredragsholderens eget
forskningsområde.
Se i den forbindelse HyLoMOL-projektets hjemmeside (hylomol.ruc.dk).
Liana Razmerita: Ontology-based User Modeling for Knowledge Management Systems
Abstract
What are the key success factors for a knowledge management system
(KMS)? The design and implementation of successful knowledge management
systems are topical research areas. We argue that designing effective
knowledge management systems requires not only a focused view, which is
achieved by considering organizational imperatives and technological
solutions, but it also benefits from a larger perspective that considers
a user-centred design, the individual needs of the users (e.g. work
tasks, responsibilities), individual motivational drivers, usability and
ergonomics issues. This talk emphasizes the role of user modeling within
Ontology-based Knowledge Management System (OKMS), integrating a highly
interdisciplinary approach. It shows how user models, models of the
knowledge workers, and user modeling processes can be applied in the
context of knowledge management systems. This talk emphasizes the
importance of using ontology-based representations for modeling the
users, providing enhanced user support and advanced features in KMSs.
Pernille Bjørn: Challenges Managing Groupware in Global Virtual Teams
Abstract
It is not enough just to have the 'right' kind of groupware technology
supporting collaboration in virtual teams to have success.
Managing groupware technologies in global virtual teams is
viewed as a process of integrating technology and collaboration.
This involves a continual negotiation of the team's goals, processes,
and technology. The presentation will present an investigation of
organizational factors constraining this integration process, by
analyzing the failure of integrating groupware into two global virtual
teams within industry. I will present an empirically driven interpretive
case study conducted in a large distributed global organization of around
100.000 employees examining the possibilities for integrating groupware
in two global virtual teams. Based on the empirical observations, we
reveal two organizational factors challenging the integration process:
The importance of joint enterprise and the role of the intermediator.
Mads Freek: AJAX - or how to make web applications truly interactive?
Jesper Simonsen and Morten Hertzum: A Regional PD Strategy for EPR Systems:
Evidence-Based IT Development
Abstract
One of the five regions in Denmark has initiated a remarkable and
alternative strategy for the development of Electronic Patient Record (EPR) systems.
This strategy is driven by Participatory Design (PD) experiments and based on
evidence of positive effects on the clinical practice when using EPR systems.
We present this PD strategy and our related research on evidence-based IT development. We report from a newly completed PD experiment with EPR in the region conducted through a close collaboration comprising a neurological stroke unit, the region's EPR unit, the vendor, as well as the authors.
Dan Corbett:
Interoperability of Knowledge Bases
Abstract
Traditionally, knowledge-base research has centered on how to capture,
archive and index knowledge, so that it can be used to solve future
problems of a similar nature. The problem facing us today, however, is
not in storing the knowledge, but in sharing, indexing, and
disseminating the knowledge, and then using the knowledge to collaborate
on solutions.
Further, we must finally come to terms with the fact that all knowledge
is created from a specific perspective, or point-of-view.
The goals of our research have been to define an architecture for
knowledge systems that allows the knowledge not only to be shared, but
also allows different perspectives to be represented. To this end, we
have designed and built an ontology extraction system that can be used
to create a light ontology for any knowledge domain. In this seminar,
we will explain the techniques used, and show how the system can be
applied to problem domains.
Henning Christiansen:
Automatisk analyse af tegnsprog: Om baggrund og metode for et nystartet
forskningsprojekt
Abstract
Dette projekt handler om døves tegnsprog.
Hvor automatisk analyse af skrevet og til dels talt tekst er en
veletableret disciplin med udbredt teknologi, er det helt anderledes
med tegnsprog, i særdeleshed dansk tegnsprog. Der foreligger stort
set intet; det som kommer nærmest er et iøvrigt meget spændende
ordbogsprojekt, som vi har kontakt med. Dette projekt vil
forsøge at etablere et grundlag for en sådan analyse, om muligt
med afprøvning i en prototypeanvendelse.
På dette seminar vil vi forsøge at indfange hvad der
er specielt ved tegnsprog, og forklare, hvorfor det netop er os,
som går i gang med dette projekt.
Gourinath Banda:
CLP-based semantic modelling to reason about Component-Based software designs
Abstract
There is often a mismatch between the structural and functional specification of
component-based systems. The structural description shows how functional blocks
are connected together, but gives no explicit information about the flow of
control or the state of the system at a given instant. In this abstract we show
how the structural description of a component-based design can be represented as
a logic program and automatically transformed using standard logic program
transformation tools into a behavioural description, namely a state-based
transition system. Each description can be checked using logic-based analysis
techniques. For example the interfaces between components can be checked using
type inference methods, while safety properties and other dynamic aspects can be
checked by applying static analysis techniques to the behavioural description.
We illustrate the approach using the component-based embedded systems design
framework COMDES.
Slawomir Zadrozny:
Expressing preferences in queries: some interpretations
Abstract
Techniques are presented for allowing the expression of certain types
of conditions in queries in a more
intuitive way. The key concepts are bipolar queries and queries with
preferences. For example, in case of the bipolar queries the user may
differentiate between the required and preferred conditions. This may
be exemplified by a query: "Find cheap houses, preferably near the
railway station". Various interpretations of the queries of this type
are discussed. The queries with preferences were defined by Chomicki
together with a new relational algebra operator called "winnow". We
propose the fuzzy version of the winnow operator and use it to
express a selected interpretation of the bipolar queries.
John Gallagher, Kim Henriksen and Gourinath Banda:
The ASAP Project: Conclusions and Future Work
Abstract
The ASAP project (Advanced Specialization and Analysis for Pervasive Systems)
was an EU Framework 5 project running from 2002-2006, with partners in Spain,
UK, and Germany as well as Roskilde University. In this talk we look back at
ASAP, summarise some of the project results and ask to what extent it achieved
its goals. We focus on the part of the project carried out at Roskilde and
describe current research that emerged from ASAP. In particular we look at
a tool-kit for logic program analysis and specialization, available on the
web, and show how it can be adapted to analyse code in other languages.
Jan Pries-Heje:
Situationsbestemt valg af udviklingsmetode
Abstract
Siden de allerførste systemudviklingsmetoder blev beskrevet, har vi
diskuteret hvordan man skulle vælge hvilken metode der skulle bruges.
Talrige bøger (og konsulenter) har hævdet at være i besiddelse af de
vises sten, men vi erkender i dag at ingen enkelt metode nogensinde
vil kunne bruges i alle situationer. Det bedste vi kan gøre, er at
blive i stand til at vælge mellem de egnede systemudviklingsmetoder
i en konkret situation. Dette foredrag præsenterer resultaterne fra
et forskningsprojekt, hvor vi dels har udviklet en konkret "stjernemodel"
til at karakterisere projekter og fokusere valget af metode, dels har
arbejdet sammen med Danske Bank gennem de seneste 4 år om at udvikle
en tilvalgsmodel, der giver anvisninger på fornuftige valg gennem
labyrinten af mulige kombinationer af metoder og teknikker til
systemudvikling.
Magnus Nilsson:
Representing time in calendar and scheduling systems - on scheduling and temporal coordination in home-care work
Abstract
In calendar and scheduling systems time is represented in terms of clock time.
Such a representation facilitates an objective, quantifiable, and precise
measurement of time. Clock time, however, cannot fully portray peopleÕs rich
and diverse means of accomplishing temporal structure and alignment. Based on
an ethnographic study of home-care work, I will discuss the affordances and
constraints of clock-time specifications in calendar and scheduling systems.
Further, I will illustrate how other types of temporal conceptualizations -
for instance, rhythms, temporal horizons, and temporal boundary objects -
might add to a more adequate understanding of temporal alignment and
coordination among cooperating ensamples. Finally, drawing on the ethnographic
study of home-care work, I will consider the relation between objective
clock-time specifications and more people-based conceptualizations (e.g.
rhythms, temporal horizons, trajectories).
Ada Scupola:
Factors affecting e-commerce adoption in Danish and Australian SMEs
Abstract
This paper presents the results of a study investigating environmental, organizational and technological factors influencing adoption of electronic commerce in small and medium size enterprises in Denmark and Australia. A total of eight companies have been interviewed, four located in the area of Copenhagen, Denmark and four in Brisbane area, Australia. The results show many interesting similarities and few differences between the two countries. For example, in both countries the external environment has an influence mainly through customer pressure and availability of ICT consulting services. The main organizational factors affecting e-commerce adoption in the two countries are the CEO and employees' knowledge and attitude. Finally a number of technology related factors affect e-commerce adoption in the two countries as for example the ability to stream information out to the clients or the possibility to offer online training.
Full paper
Keywords: Adoption, e-commerce, Internet, SMEs, Denmark, Australia
Keld Helsgaun:
Improvements of the Lin-Kernighan heuristic for the traveling salesman problem
Abstract
The Lin-Kernighan algorithm is generally considered to be one of the most effective heuristics for solving the traveling salesman problem (TSP). The algorithm is specified in exchanges (or moves) that convert one candidate solution to another. In this talk I will present improvements of the algorithm and describe an effective implementation of k-opt moves.
Niels Jørgensen:
Eksemplarisk designpraksis
Abstract
En af udfordringerne i forbindelse med det nye basisstudium, humtek, er
at der er knyttet mange forskellige fag og fagligheder til studiet. Det
rejser spørgsmålet om de studerende udbytte bliver sammenhængende eller
fragmentarisk. Eksempelvis kan man spørge, når der deltager undervisere
både fra datalogi og geografi, hvor stor sammenhæng der er mellem
software-design og byplanlægning (urbant design)? På seminaret vil jeg
diskutere begrebet 'eksemplarisk designpraksis'. Kan dette begreb, der
er parallelt til begrebet 'eksemplarisk projektarbejde' fra den
pædagogiske teori om problemorienteret projektarbejde, indfange et
væsentligt aspekt af læringsprocesserne på et designorienteret
baisstudium?
Sisse Siggaard Jensen:
Sense-making og brugerdreven innovation i Virtuelle Verdener
Abstract
Avatarer er de figurer, der befolker de virtuelle 3D verdener. De er nye medier for kommunikation, og de findes i mange forskellige former. En væsentlig skelnen er den, der går imellem programmørens/ designerens henholdsvis aktørens rolle i forhold til det nye medie.
Det aktuelle forskningsprojekt: Sense-making strategies and user-driven innovations in Virtual Worlds, 2008-2011, er støttet af Det Strategiske Forskningsråd, KINO. Det bygger på tidligere projekter.
I mit oplæg vil jeg berette om det aktuelle - og også meget gerne om de forudgående projekter.
Troels Andreasen:
Conceptual Summaries as Query Answers
Abstract
This talk presents an approach to retrieval where conceptual summaries
are given as answers to queries. The idea is to restrict a general world
knowledge ontology to a given set of concepts and thereby providing a
structure, a so-called instantiated ontology, for navigation and further
investigation of the concepts. Typically the restriction will be to the
concepts appearing in a set of documents or an entire corpus.
In this talk we are specifically concerned with the instantiated ontology
as source for extracting summarizing concepts characterizing query answers.
