NWeSP 2007 review - day two

Day two of the conference started with a key note, presented by Kaori Yoshida and Mario Koeppen: "The Gestalt of Web Services". A very interesting presentation of visual aspects of web design. Unfortunately it was mostly off-topic for this conference. Maybe not if there was a standard visual representation for a set of web services and "aesthetic" metrics to "see" if the services looked good or bad (ex. too many dependencies)...

Hridesh Rajan working in Iowa State university, USA, presented the paper "How to Trust Web Services Monitor Executing in an Untrusted Environment?". It was one of the most interesting papers of the whole conference. It proposes adding hardware-based trust verification capabilities to SOA, using TPM - Trusted Platform Module - embedded in the servers' microprocessors (ex. Intel's). Only the trust monitors are bound to the server hardware TPM but not the services executing on that machine. This is an interesting approach to build a TCB - Trusted Computing Base - in a distributed system. However, there has to be a hardware registry of the servers, and a trusted procedure for the distribution of public keys when new servers are added to the infrastructure.

Another very good paper was presented by Gottfried Vossen, from Germany: "Web Service Discovery - Reality Check 2.0". The paper discusses the real world use of Web Service registries, like UDDI. The main conclusion is that UDDI may be interesting in a closed world inside an organization (especially in a single vendor environment - SAP, Microsoft, IBM, ...), but in the Internet, service discovery is mostly a human-centered process: people find out about the service and link directly to its contracts. The Universal Business Registry vision might never come true.
On a side-note, the paper also does a good Web 2.0 overview, highlighting its main principles and differences to Web 1.0.

The paper "ATHENE - An Approach for Modeling Semantic Web Processes over User-friendly and Editable Ontology Models" by Nishant Singh from India, presents a well structured approach to semantic description of services, using a user-driven process, with a three layer model. The users are developers and/or domain experts. The tiers are Meta2-model (object, types and relations), Meta-model (using OWL) and Model. There are also some user-interface abstractions - like notebook - to help the users edit the ontology.

Next, I presented my work "Core mechanisms for Web Services extensions" (in case you don't remember, I'm Miguel Pardal from Portugal's Instituto Superior Técnico). I described the core mechanisms necessary to build Web Services extensions, regardless of the underlying platform. I also mentioned the on-going work to build an extensions engine integrated in a three-layered application framework. In the Q&A, I talked about general-purpose extensions (ones that add some specific funcionality or a WS-standard implementation) and customization extensions (ones that add features to an existing capabilities). A good analogy for WS-extensions are Mozilla Firefox's extensions.


The next presentation was "A Framework for the Requirements Analysis of Service-oriented Workflows" by Jochen Muller, from Germany's University of Kaiserslautern. The paper presents a workflow definition where it is a service and is composed of (stateless) services. The methodology distinguishes between process design and enactment stages of the workflow. The workflows can be long-running, with some instances taking years. The system is focused in the e-Government domain, particularly, in the gov-to-gov interactions. In my opinion, this proposal would benefit from having an explicit account for versioning, in order to have different version workflows running side-by-side.

The final presented paper was "On The Use of Ontology Reconciliation Techniques in SOA" by Patricio de Alencar Silva, from Brazil. This work is about the Semantic Web and its automatic service discovery and binding. The paper acknowledges the fact that semantic heterogeneity is not a computational problem, but a philosophical one. It presents the existing approaches to brigde concept definitions gaps between different organizations and application domains using ontology-based techniques.

NWeSP 2007 review - day one

I'm now going to present some highlights from NWeSP 2007, the 3rd International Conference on Next Generation Web Services Practices, held in Chung-Ang University, Seoul, South Korea, from October 29 to 31 , 2007.


In the opening session, Dr. Nikolay Shilov from Russia, presented the talk "Combining logics of Knowledge, Time and Actions for reasoning about Intelligent Agents". It was about the different approaches to knowledge: formal concepts analysis (FCA), description logic (DL) and other logics of knowledge. The first two are suitable to represent static facts about objects and their relations. The latter deal with uncertain and evolving environment acquisition. Dr. Shilov argues that these are the most suitable to build Semantic Web Services.
It was a good talk. My main criticism is that the present results are too distant from current Web Services applications.

The first presented paper was "A Simple Approach to Optimize Web Services Performance" by Tanakorn Wichaiwong from Thailand, about the performance improvements of using FTP transport for Web Services (WS-FTP) in an Intranet environment comparing to HTTP. Security issues were not analysed.

The next presentation was "Transforming valid XML documents into RDF via RDF schema" by Thuy Pham from Korea, about the conversion of XML-DTD data structures to RDF format, more suitable for Semantic Web, where the information is represented in triples: subject, predicate and object.

Next was "First approach to Design a Web Service Honeypot: WS-pot" by Hugo Gonzalez from Mexico. Hugo described his system that applies the honeypot approach to web services security: simulate a service and attract attackers to record how they try to compromise the service or the server machine. Promising work, especially if the real-world attacks data will be related to the effectiveness of security mechanisms provided by the WS security standards.

The next two papers were presented by Dimitru Roman, working in Austria.
The first one was "On describing and ranking services based on non-functional properties" about a description and automatic binding mechanism for Semantic Web Services. The presented work was tested with lab application. It is a gradual phased-approach to Web Services, that might be an interesting approach. Check http://www.wsmo.org/ for more information. In my view, the greatest problem is that it assumes uniform knowledge (ontology) between the cooperating organizations, which is difficult to achieve in real-world applications.
The second one was "A CTR-based Approach to Service Composition Patterns" describes the use of CTR - Concurrent Transaction Logic (only atomicity property) - to implement different workflow patterns. This paper is based on the patterns catalogue available at http://www.workflowpatterns.com/.

Next "The WINGS of jCOLIBRI: A CBR Architecture for Semantic Web Services" was video-presented by Juan A. Recio-Garcia, from Spain. CBR stands for Case-Based Reasoning and proposes a client-server binding architecture based on a history of previous interactions (cases), finding similarities with current invocation and performing some minor adjustments. The authors find limitations in their proposal that they will attempt to solve using planning techniques.

The final presentation of the first conference day was "A Framework for Measuring Performance in Service-Oriented Architecture" by Si Won Choi, from Korea. It was a very good paper about Web Service performance metrics. The main benefit of the proposed metrics is their practical measurement procedure. The metrics were obtained with a statistics collector and a performance monitor for a hotel reservations web service, built on top of the Apache ServiceMix ESB.

The most interesting poster was "Aspect Mining for Dynamic Service Orchestration" by Hao Yang, as it presented a practical application of AO4BPEL, an aspect-oriented extension for BPEL.

Fun and study / Diversão e estudo

"In Portugal, the education policy of the last decades has been moving in the wrong direction. Instead of going from entertainment to education, it has gone from education to entertainment. Instead of promoting fun toys and games that (discreetly) also teach and develop abilities; it has promoted fun study, where what matters is trying, not achieving, with a loss of responsibility and individual commitment, whose negative effects we feel today in the lack of excellence in work."

~ Joana

/ This post is bilingual: English (U.S.) and Portuguese (Portugal) /

"Em Portugal, a política de educação das últimas décadas foi na direcção errada. Em vez de se ir da diversão para a educação, foi-se da educação para a diversão. Ou seja, em vez de se promoverem brinquedos e jogos divertidos que (discretamente) também ensinam e desenvolvem capacidades; promoveu-se o estudo divertido, onde o que interessa é o tentar e não o conseguir, com perda de responsabilidade e de empenho pessoal, cujos efeitos negativos se sentem hoje na falta de excelência no trabalho."

~ Joana

In line / Na linha

I've been in life's line for long enough to know it's not a straight... / Estou na linha da vida há tempo suficiente para saber que não é uma recta...

Useless

Someone special sent me this: / Alguém especial enviou-me isto:


In love's math, the unit(y) is infinity... / Na matemática do amor, a unidade é o infinito...

Why publishing matters?

Some of you might ask: why so much trouble finding about conferences or journals? Can't I just do my job, get my degree and get on with life? Do I really need to write a paper about what I'm doing?

In my experience, publishing is more important than it seems at a first glance. Why? Mainly for two reasons:

1 - When you write down your ideas and work in paper form - with concise, well-written text, and full referencing - you are forced to summarize your ideas and to look at what other people are working on. It makes on focus on what you do better or differently. It makes you distill what you did.

2 - When you share your work with an interested audience, you get feedback from other people that can reason about what you're doing. This enables you to anticipate the criticisms your work will face, and do something about it before it's finished.

Conference posters are more suitable for on-going work, as they maximize the feedback from more diverse people. Conference paper presentations are usually attended by fewer, less diverse people, but the feedback is more focused.

Never forget to make your contacts visible! Sometimes the feedback comes after the conference.

You can also keep a publications page, so that people may easily find your work on the web. However, there must be a disclaimer.

Sharing International Conferences with Google Calendar

Following my paper search, I identified international conferences on subjects relevant to my research. These are potential submission targets. Papers are excellent deliverables that can also work as a research milestones.

I use a public Google calendar to share this information with anyone who is interested. The address is the following: International Computer Science Conferences

If you want to cooperate in keeping this calendar up-to-date, please let me know.