The second major task for Ingo was the finalization of the source code packages. He created two different packages: the Generic Connection Server (GCS) package and the Connection Server (CS) package. The CS package includes the source code of the Connection Server that was used in the Itchy Feet project to create the 3D e-Tourism environment. Additionally, this package contains several automated tests that simulate the 3D Virtual World and which can be executed in combination with the release package of the 3D e-Tourism environment.

The Generic Connection Server which is available in the GCS package is a stripped down version of the Itchy Feet Connection Server with all the Itchy Feet specifics removed. This package is intended for developers who want to utilize our framework for the development of custom Virtual Institutions. The GCS package also includes an Example Federation and automated tests that simulate a 3D Virtual World. This makes it possible to execute the GCS, to run several tests against it and to learn abut and understand its functionality.

Both packages are accompanied with an extensive, Wiki-based, documentation that provides information on the background of the framework, setup and execution instructions as well as implementation details. The packages can be downloaded as zip or tar archives in the Download section.

Finally, we did receive some good news on our submitted book chapter on the Hybrid Semantic Search System (HS3). The editor did send us two positive reviews (with a third pending) and gave green light to prepare the final version of the book chapter.

Ingo’s work in the Itchy Feet project has ended now and he will soon be leaving for Ireland where he is going to spend the next year together with his girlfriend in the city of Limerick. The last three years have been a valuable experience and it was a great pleasure to work together in a team with such highly skilled people as Helmut, Markus, Josef and Michael. He will be missing the very enjoyable working atmosphere and already looks forward at an Itchy Feet team reunion in the future.

Last week Markus adapted the Data Fetching Routines of HS3 and will try to improve the Annotation
Component of HS3 to get better results this week. Furthermore, he’ll create an example corpus of approx. 100K documents and 800K triples to evaluted the scalability of the system. He visited Andy Rauber to give him a presentation of HS3 and got all documents that are needed to officially register his PHD thesis.

The last weeks Ingo has been working on the creation of a generic Connection Server package which can be used by other researchers to implement Virtual Institutions. In particular, it is intended to support Tomas in the connection between Electronic Institutions and 3D Virtual Worlds in his research domain of automatically creating 3D Virtual Worlds out of Electronic Institution specifications. While the static generation has already been realized by Tomas, the generic Connection Server package will aid him in making the environment dynamic.

Furthermore, Markus and Ingo reviewed two book chapters for the book “Ontology-Driven Web Mining”. We had received these review requests as we had also submitted a chapter for publication in this book, and, apparently, the editor had run out of reviewers and kindly asked us to also review two chapters. And this we did.

Ingo is also eagerly awaiting his Rigorosum (comparable to a PhD defense) which will take place tomorrow on Monday, June, 21th. He has prepared a talked which he already had presented in front of the members of the VSEM project this week and is currently working on further refining and optimizing the talk in order to give a good presentation tomorrow.

The meeting took place on Thursday and was divided into three different parts. In the first part, Ingo gave two presentations – one on the Itchy Feet system and one on the evaluation of Itchy Feet – and Tomas gave a presentation as well as a demo of the shape grammar interpreter (which is the graphical tool for the design, specification and generation of the virtual world). Then, in the second part, we gave them an in-depth demo of the Itchy Feet system showcasing all the available functions and illustrating how the connection to the Multi-Agent System works. In the last part, which was the most extensive part, Markus and Ingo introduced Tomas to the package structure and source code components of the framework. We showed him how the connections between the different layers are established, how the fraemwork is controlled and which kind of functionality is implemented in which package.

In conclusion, the meeting went very smoothly and it was a great pleasure to have the two guys here for one day. Furthermore, we did get quite a lot of information exchanged and we believe that Tomas is now ready to experiment with the framework and to possibly utilize it for their work.

The last two weeks have been quite busy for Ingo. Especially the last weekend was stressful as he worked hard on his dissertation to be able to submit it on May, 18th. He got everything finished in time and then submitted the dissertation on Tuesday. He is now awaiting his PhD defense which will take place in mid June.

Markus wrote his PhD proposal, which was initially intended to be a 1 page summary but finally grew to the size of 10 pages. Furthermore, he adapted the GWT-based UI. Now the input field dynamically adapts to the size of the input, it is possible to input names in double quotes if they contain whitespaces and it’s possible to use the AND keyword to include multiple restrictions for a recognized concept.

HS3 is currently in a pre-alpha state. It can be used with the Itchyfeet knowledge base and Harmonise ontology. Markus is currently refining the web-based input mechanism and the layered search functionality. Within the next 4 weeks the bugs discovered during testing will be fixed, the performance of the system improved and an architecture cleanup done. Subsequently, the IE mechanism will be enhanced to reduce noise in the document-based knowledge base which is build by automatically fetching domain related documents off the Web. The enhanced IE mechanism will be used to generate the final document-based knowledgebase which will be used to perform a user-centric evaluation of the system and web-based interface. Furthermore, the components (i.e. IE, Annotation, Indexing, IR, Ranking) of the system will be evaluated separately as well. Finally, the architecture of the system, the developed mechanisms and evaluation results will be published in an appropriate journal.

Josef finished a first version of the ICURA paper and was prepared to submit the paper on May, 6th. However, as the paper submission deadline for the VSMM Conference was extended to May, 26th, there is now still enough time to further revise and improve the paper. The next days he will think about ways to integrate the findings of the ICURA project into the VSEM project. Ingo was intensively working on his PhD and did submit a final version to his supervisor on May, 2nd. If everything goes well and there are only minor corrections to make, he will submit his thesis at the next submission deadline on May, 18th.

From the beginning of May Josef will change to the project “The Virtual 3D Social Experience Museum” at Vienna University of Technology. So the times at Matrixware are nearly over for him. During the remaining time Josef will continued writing the paper about ICURA for the 16th International Conference on Virtual Systems and Multimedia (VSEMM 2010). He is making good progress and will send out the paper for a first review cycle during the next week.

The paper about ICURA will be submitted to the International Conference on Virtual Systems and Multimedia (VSMM ‘10) – Website – Deadline is May, 7th. A first version of the paper will be finished by the end of this week and will be sent out for revision.

Regarding HS3, the indexing mechanism was extened to index the occurance count of entity instances per document. This information is needed to decide if a document is about a specific entity or mentions multiple entities. In the first case a entity centric search algorithm using SIREn can be used. In the latter case another search algorithm needs to be used, because SIREn can only cope with star-shaped queries and is not able to join over multiple contexts.

]]> http://www.itchy-feet.org/?feed=rss2&p=893 0 http://www.itchy-feet.org/?p=875 http://www.itchy-feet.org/?p=875#comments Sun, 11 Apr 2010 18:21:05 +0000 jfroschauer http://www.itchy-feet.org/?p=875 During his days off Josef transcribed the interviews from the evaluation. He wrote down the essential passages and stored the whole information in an EXCEL File. The file holds five different tables: Game Results, Pre-Questionnaire, Post-Questionnaire, Pre-Test Answers and Post-Test Answers.

On Wednesday Helmut asked Josef to set up an online highscore board for ICURA. He created a database that stores the nickname, number of correct answers on the Pre-Test, number of correct answers on the Post-Test, the final score and a timestamp (ordered by score, descending). The nickname is requested at the end of the Pre-Test. You can display the table here or simply follow the “ICURA Highscore button in the upper right corner.

As you can see, it shows the data from the evaluation so far. Note that the timestamp is incorrect for this data because this functionality wasn’t implemented when the evaluation was conducted. But as soon as someone plays the latest version of ICURA and finishes the post-test, the results are sent to the server. The new version will be availabe here soon, stay tuned!

This week Markus started to migrate the entire HS3 project (formerly known as Hibrid Semantic Search Project) to Maven. Maven is a Build Management Framework. The reason for this step is that it gets really complex and hard to manage due to all its different components and dependencies. Anyway, the main problem is that e.g. frameworks such as Gate are not Maven projects and do not have Maven repositories, Even worse, several java libraries used by Gate are not available in any public Maven repository so they have to be added manually to the local repository, which is not the way Maven is intend to be used. To overcome this, the proxy repository Artifactory was set up, which now acts as the central point of contact for the Maven dependency management mechanism and offers a convenient GUI to manage Maven artifacts. Furthermore, Markus read the documentation of CouchDB to figure out whether it can be use to overcome the restrictions he’s currently facing with SIREn.

Ingo was on vacation this week.

Josef held short interviews before and after playing ICURA. The game itself included two tests that show the learning success of every participant. All in all the results are very positive. On average, 5.05 (σ=1.99) questions were answered correctly in the Pre-Tests (out of 12) – after playing ICURA this value grew up to exactly 10 (σ=1.26). Two participants were able to solve all questions correctly. The average age of our participants was 29.35 years (σ=4.49). One small flaw is that our test sample is a little bit biased. The average time our participants spend on the computer per week is 40.7 hours, which is surely not a representative value. But I think, we have to live with that. The most important fact is that ICURA definitely shows a clear learning success and people had fun while playing it (Fun Factor: 4.2 out of 5).

Markus carried on working on the book chapter by incorporating Ingo’s and Helmut’s review comments. Additionally, he had a meeting with Andi Rauber regarding the book chapter and incorporated his comments during the weekend. Sunday night he submitted the chapter to the editors.

Ingo was mainly working on his PhD thesis week. Additionally, an interesting cooperation opportunity opened up this week. Tomas Trescak and Marc Esteva from the Artificial Intelligence Research Institute (IIIA) in Barcelona initiated a request for cooperation. They are working on the automatic generation of 3D Electronic Institutions via shape grammars (see here and here for a video on their work) and have expressed their interest in the middleware layer of Itchy Feet. The middleware layer connects the Electronic Institutions with the 3D Virtual World. We did have a telephone conference this week in which we talked about their requirements and they pretty much seem to fit the functionality of the middleware layer. We have now planned to extract the respective source code into a separate package until the end of June, but will provide the source code beforehand such that they can get an overview on the code structure. Additionally, they are probably going to visit us in mid-April to discuss the requirements face-to-face and to work together on this integration.