Conference Report
Applications of Logic Programming in the Semantic Web and Semantic Web Services
(ALPSWS'06)
Axel Polleres
Universidad Rey Juan Carlos
Spain
|
|
Editor: Enrico Pontelli
|
This
report gives a brief overview of the topics and results of the workshop
on ``Applications of Logic Programming in the Semantic Web and Semantic
Web Services (ALPSWS 2006)'', which took place for the first time
on August 16th 2006, in Seattle, WA, USA, co-located with ICLP.
Besides the usual presentation of scientific papers, we adopted a
methodology called Open-Space in order to facilitate a more
interactive workshop atmosphere, which, although not yet very
common for scientific workshops, proved to be very stimulating and even
led to the concrete planning of collaborations among the workshop
participants.
The advent of the Semantic Web promises machine-readable semantics and
a machine-processable next generation of the Web. The first step in
this direction is the annotation of static data on the Web by machine
processable information about knowledge and its structure by means of
Ontologies. The next step in this direction is the annotation of
dynamic applications and services invocable over the Web, in order to
facilitate automation of discovery, selection, and composition of
semantically described services and data sources on the Web by
intelligent methods; this is called Semantic Web Services. Many
workshops and conferences have been and are being dedicated to these
promising areas. They mostly deal with generic topics and
bringing together people from a broad variety of research fields with
different understandings of the topic. The plethora of these workshops
and conferences makes it hard to keep track of the various approaches
of a particular technology, such as declarative logic programming in
our case. So, the goal of ALPSWS was somewhat different: We aimed at
advancement of specific applications of Logic Programming as a paradigm
for declarative knowledge representation and reasoning for the Web. The
idea was to bring together the impressive body of work related to
applications of Logic Programming to Semantic Web and Semantic Web
Services in order to bring together people from different
sub-disciplines of LP and focus on technological solutions and
applications from LP to the problems of the Web.
Co-locating the event with one of the major Semantic Web conferences
(ISWC, ESWC, ASWC) seemed like ``carrying OWLs to Athens'' (BTW, this
year's ISWC indeed takes place in Athens... Georgia, USA, however ;-)
), so we decided to co-locate the workshop with ICLP in Seattle with
the idea to further promote research in this interesting application
field in the LP community.
The workshop took place on Wednesday, August 16th; it lasted a full day
and was divided into three sessions. While the first block was
dedicated to papers on Semantic Web reasoning, the second block hosted
contributions for applying LP to reasoning about Web services and
Policies. A final block was dedicated to discussions around the topic
of the workshop, adopting the so-called Open-Space methodology, which was started with a half an hour poster session in order to stimulate discussion.
From the seven accepted full papers (see http://events.deri.at/alpsws2006/#program),
four treated core LP and Semantic Web reasoning issues, whereas three
were mainly concerned with Web services. Among the five posters, which
were
additionally presented, two were concerned with Web services, one
with syntax, and we had two interesting contributions on OWL
Reasoning using Prolog.
Concerning adopted LP paradigms, a tendency, which could also be
observed at ICLP, was recognizable: Whereas most of the papers used
traditional Prolog, a considerable number of contributions was
dedicated to the recently emerging Answer Set Programming paradigm,
where both paradigms proved valuable in their particular applications.
Roughly speaking, Prolog-based approaches have proven valuable for
implementations of Semantic Web and Semantic Web Services solutions for
a variety of applications, as well as for building lightweight
reasoners, whereas the ASP contributions focused more on the aspects of
using LP itself as a KR formalism for the Web. It will be interesting
to follow the ongoing development, not only in the particular field of
Semantic Web, but in the LP community in general.
One of the most interesting parts of the workshop for us as the
organizers was the open-space discussion in the afternoon session.
Open-Space is a methodology for joint agenda finding (see e.g. http://en.wikipedia.org/wiki/Open_Space_Technology)
which facilitates dynamic discussions. It turned out that the
methodology is well-suited for such a workshop. The interesting thing
about this form of discussion is that you never know what happens,
since the participants themselves decide about the items to be
discussed, and several discussions are happening in parallel in
the same room; people join and leave discussions absolutely
voluntarily, but it (surprisingly?) still almost always works!
Six concrete topics were discussed throughout the Open-Space sessions
which we describe (in a deliberately sketchy way) in the
following. Not all discussions had a ``tangible'' outcome, but
many interesting issues were raised which increased the common
understanding of the problems in the area.
Open world reasoning vs. closed world reasoning.
This topic was concerned with the different implications and needs for
Semantic Web Reasoning. Much time was spent on reaching a common
understanding of the distinction between "open" and "closed" in the
context of the Web. E.g., negation as failure may be viewed as closed
world reasoning on the one hand, but also be interpreted as simply
reasoning by default, which is very natural for humans. OWL, as opposed
to LP dialects advocates a strict open world reasoning. However, if the
Web is to be the ``world's largest large database'', what are the
implications, as database query languages (including SPARQL!) do have
negation as failure? Closed world reasoning is very common in
databases. Can scoped negation over closed Web documents or local
closed world reasoning do the trick?
Ontologies and Web Services.
This topic was concerned with the meaning of Ontologies as such, as
well as the meaning of Ontologies for (Web) services. The focus,
though, was on the former, including the discussion of different
Representation Formalisms and different Ontologies. Other topics: What
does it mean to ``harmonize'' different ontologies? How to do ``Web
services search''? In how far do/can search engines make commitment to
``meaning''? Reuse as a crucial issue of ontologies; discussion of a
concrete application area - eHealth - where standards with
completely different backgrounds and communities exist.
Smart Search Engines.
Challenges and open issues regarding smart search engines were
discussed: How to reduce annotation costs which are a high entry
barrier to the Semantic Web? Which technologies can be used? How to
increase the benefit compared to current search engines? What
\emph{output} do you expect from "smart" search engines? What is the
meaning of correctness/reliability/locality in search results
(correctness vs. "google")? How can you specify constraints?
Semantic Modeling of Web Services Inputs/Outputs.
This slot was dedicated to the often overlooked problems of
semantically modeling inputs and outputs of Web services. It is often
not enough just to say that the input shall be an instance of an (OWL)
ontology, because this raises the following question: What to send over
the wire? CORBA and XML Schema both provide formalisms to check valid
inputs, whereas this problem is to a large extent ignored/neglected in
Semantic Web Services frameworks. It remains unclear how to set up a
contract, when e.g. an instance of the class "Person" is requested as
service input. What shall be sent by the requester? A person id? A name
and social insurance number? It was agreed among the workshop
participants to jointly work on a solution of this issue, defining a
formal specification of semantically described input and output.
Joint project opportunities.
Which possible funding lines for joint projects exist on EU level? How
can US (or other non-EU) partners join? What are the most
promising/challenging application areas (also from the perspective of
funding possibilities)? Several participants showed interest in joint
projects and first contacts have been made (and, in the meantime, have
even led to the first joint proposals).
Next ALPSWS Fortunately,
everyone agreed that we had a stimulating and useful workshop, which
made positions, approaches, but also open issues and gaps clear. One of
the main issues turned out to be the question how to bridge the
gap between Semantic Web and Web services. LP methods are successful
for both, but so far hardly in a unified framework: In this slot, we
discussed where and when to have a next ALPSWS in order to
further encourage cross-fertilization of these fields.
We further discussed how to keep the spirit to make the event a workshop
instead of a mini-conference. The Open-Space discussions were agreed to
be a perfect means to this end. Additionally/alternatively,
possibilities such as a panel or different forms of presentations (e.g.
10min flash presentations plus a longer poster session for all papers)
were discussed.
Summarizing, the first three of the Open-Space sessions rather served
to achieve a higher level of mutual understanding, than leading
to specific joint work, whereas the latter three already seem to
have sparked concrete activities.
We plan to collect the outcome of these discussions on a separate Wiki
page (which will be made accessible via the workshop Webpage (http://events.deri.at/alpsws2006/)
soon and hope that the participants, but also other interested people
joining, will take up the defined action items and be able to
present more concrete answers to the open questions mentioned
above.
Proceedings including the full papers and abstracts of presented
posters are available online in the CEUR Workshop
Proceedings series, under Vol-196 (see http://ceur-ws.org/Vol-196/).
Additionally, the two best papers will be published in a dedicated
section of the upcoming special issue on "Logic Programming and the
Web" in TPLP, see http://www.dsi.unive.it/~tplp/.
Hopefully, see you -- and awaiting your interesting contributions - at a possible continuation of this workshop!
Axel Polleres, Stefan Decker, Gopal Gupta, and Jos de Bruijn (the organizers)