To continue with the SOAP::Lite issues mentioned in the last posting: the problem is solved. My recommendation: NEVER EVER use ‘encoding “utf8″‘ when you want to deal with utf8. Instead, use ‘Encode’ and convert all the strings manually. Tsssss, perl and unicode don’t go well together. Anyway, the batch upload is now working, we’ll clean up the stuff and then provide it to everyone who is interested!

After the maybe useless but funny spherical tag cloud introduced last week, we are proud to have yet another fancy and interactive eyecatcher in ALOE: The “Resource Flow View”.

This view was realized with a slightly adapted version of ImageFlow 1.0.2 by Finn Rudolph. It won’t load images dynamically during navigating, so it loads all thumbnails in advance - this might take a while. We’re still experimenting a little bit with this stuff, so it’s not offered as a menu item in ALOE. To see it, you have to add the parameter “viewMode=flow” on any page where resources of a certain user are shown (this also includes “MyResources”).

Last but not least, we found some bugs in our Live System that we should fix pretty fast. E.g., changing the sort order for results can result in strange behaviour, because certain parameters just disappear… Depending on how long this will take, the release of the new OPENEER might be a bit later than expected.

Procrastination is of course bad when trying to finish a PhD, but it’s good when it’s about new features for ALOE So we now have a nice, fancy and interactive “spherical tag cloud” to display popular tags in ALOE. I’m not sure whether it’s really useful, but who cares, it’s fun and it’s definitely an eye-cather! The spherical tag cloud was realized with a repurposed version of the great WP-Cumulus WordPress plugin by Roy Tanck - check out his blog for more information on how to repurpose his code.

Furthermore, we’ve put a lot of effort in the creation of an install script that allows to deploy instances of ALOE in a more convenient way. Setting up the different components and setting the parameters manually was always quite a lot of work. It runs smooth now, so the updated OPENEER instance should be available this week!

What we also did was to create a small tool that allows for the batch contribution of resources defined via bibtex (or, to be more precise: via BibTeXML that was created with the JabRef reference manager). It’s done with our perl SOAP client, and once we found out why SOAP::Lite doesn’t want work with parameters defined in UTF-8 (everything works fine until you dare to use something like an “ä” or “ü”), we’ll be really happy. Hints are VERY MUCH welcome!

Alas, it is done now. After lots of work in the last months, we today released the new ALOE! All data from the old system was copied to the new one, so don’t worry, nothing is lost. We’ve sent an email to all ALOE users with according information. When everything runs smooth now, we’ll also move the other ALOE instances such as OPENEER very soon!

Ok, what happened? ALOE might still look very similar to you, but lots of things changed, especially in the backend. Let’s start with some changes and new functionalities you will probably notice immediately:

• The Web Interface got a redesign.
• Enhanced navigation and sorting possibilities for several lists (e.g., MyResources ).
• The login is now the nickname and no longer the email address.
• We added a “Forgot your password?” functionality.
• Atom Feeds are provided for any open group in ALOE.
• Messages to groups can be sent.
• A bookmarklet to add new bookmarks to ALOE is now available.
• ALOE provides a message sentbox, and ALOE messages can optionally be forwarded to your email account.

And here some additional technical information:

• The new ALOE Web Service stores resources, user-defined metadata sets and metadata about the resources in a MySQL data base. That way fine grained access control to the data which is stored in the system can be guaranteed.
• ALOE has a new JSP interface which allows URI driven navigation. The implementation is performed according to the J2EE Pattern FrontController in combination with the Action Pattern in order to make the component easy to maintain and simple to extend.
• Ajax functionality is integrated using the well-tried Action/FrontController pattern combination as well.
• Many functionalities are implemented using the taglib library in order to avoid code replication. This also comes in handy for the server side components of Ajax functionality, where responses consisting of HTML snippets can reuse the code inside the classes of the taglib library.
• We did our best to protect the platform against code injection or cross-site-scripting (xss).
• Most parts of AloeView make consequent use of the formatting tag library to enable instances of the system in different languages (or even language switching at runtime later on). Multi-language support is not perfect yet, since there are some known (and definitely several unknown) code snippets, where text was directly inserted into the pages. Nevertheless most of the functionality can be switched to a new language by simply translating one text file.
• The AloeView system validates as XHTML 1.0 Strict, although there might be some system states, where pages do not validate in all cases. It is simply impossible to check all possible combinations of database states (e.g. empty resource sets), so it is likely that validation will fail in some (rare) cases. Additionally it might happen of course, that new errors are inserted into the system since it is work in progress, but we will keep an eye on it and will do our best to guarantee standard compliance.
• We do not longer rely on an external thumbnailer for web pages. Instead we are using our own thumbnailer based on khtml2png. There is still some trouble (e.g., with flash content and when adding a bookmark that points to a pdf or jpeg), but we’re working on this.
• A new architecture of the ALOE Web Service has been implemented which provides the following advantages:
  • Improved interface in particular for JAVA clients. For such clients many methods return beans so that a confusion of return parameters can be avoided.
  • The business logic is encapulated in a seperate layer in order to enhance the extensibility of the system.
  • Encapsulation of the data base access in dedicated classes so that the exchange of the persistance layer will be eased for future scenarios.

Of course we are always happy about your feedback! Tell us when something is not working, when you’re missing something, when you don’t like certain features etc. Of course, we’re also happy when you tell us that you like the new ALOE Have fun!

You’re running the system on the same server for months. Nothing ever happened. You are testing your code. Extensively. Then you launch.

And what happens 12 hours later??? A RAM defect on the server. That’s really unfair…

Thanks to David, we could replace it quite fast, so C-LINK was “only” down from 10 to 10.45 am. Enough hardware trouble, now we hope to have some more fun with the system

After lots of coding, extensive testing (thanks a lot to all the beta testers in the KM group of DFKI!) and unexpected last minute problems (as expected, of course) we launched C-LINK today!

C-LINK is a web based tool that allows attendees of the KI 2008 conference to organize their conference.

Amongst others, features are:

• Individual conference planner (create your personal conference schedule)
• Personalized recommendations of presentations
• Meet the participants (profile matching with regard to research interests)
• Chat, Whiteboard
• Sharing digital resources (documents, presentations, etc.) among participants

As you can imagine (why else would you find this post here!?), it is based on ALOE. For the recommendations, DynaQ is used in a combination with myCBR.

We’re very excited and look forward to what happens now!

Shame on me, I didn’t blog for such a long time, although (or maybe: because) such a lot of things happened. E.g., we are partner in
the European project MACE for several months now - they use ALOE as a social layer to allow commenting, rating, tagging, and personal collections of resources. Furthermore, ALOE is used as the basis for the C-LINK system (a web based tool to support conference attendees) that will be released in two weeks. The refactoring is almost done, so a new ALOE version will be available very soon. And so on, and so on, and so on…

In the moment, I’m attending the TRIPLE-I in Graz, where I will present some work done by Rafael Schirru, Michael Kockler and me. It’s called “Providing Multi Source Tag Recommendations in a Social Resource Sharing Platform”. Slides will be available soon, as well as detailed information about MACE, C-LINK, the refactoring…so stay tuned!

Rafael and me are on tour promoting ALOE - he is presenting the paper “Towards the Combined Use of Metadata to Improve the Learning Experience” (authored by Martin Wolpers, Elia Tomadaki, Rafael and me) at the ICALT in Santander, while I am now in Vienna at ED-MEDIA, soon presenting “An Approach to Enable Collective Intelligence in Digital Repositories” (authored by Martin Wolpers, Elia Tomadaki, and me).
Both papers deal with how ALOE can be used as a social layer on top of existing infrastructures, allowing to aggregate metadata generated in various ways.

Exciting times ahead! Tomorrow we leave to Berlin for this years re:publica. Giving talks, hanging out with the German digital intelligencia and having serious fun. An excellent opportunity to attract the early adopters to try out OPENEER. The first Open Collection for Extrem Emotional Reactions to Music, a Web2.0 style platform gathering personal experiences with music on a worldwide scale. Why? If you think about designing and implementing the next generation Uber-recommenders you encounter a major bottleneck: There is few knowledge about the reasons for chills, thrills, strong emotional reactions when listening to contemporary music including the alignment of personal episodes to the lyrics. Although some good research in musicpsychology has been undertaken to decode these effects with respect to structural features in classical music, we still miss a large scale knowledge base.

With OPENEER we want to offer the possibility to volunteers to share their personal experiences in order to enable research based on this groundwork. Call it crowdsourcing or collective intelligence, we dont mind, we GO FOR IT!

OPENEER is here! Play around with it!

http://openeer.dfki.de

It’s always nice to go to Berlin, and when it’s because of the re:publica, that’s even better! The re:publica is a conference for bloggers and other net citizens, and this year’s motto is “Die kritische Masse” - “The critical mass”. The ability of the mass to contribute with criticism is a very interesting topic, of course also in the context of ALOE. The quality of metadata will become more and more important in what is often refered to as the “Web3.0″. How can this metadata be genreated? Do we trust the generators? How can we contribute with different opinions? I will also talk about this in two presentations on Thursday and Friday, slides will be available when I’m back from Berlin!

Last Friday, our intern Martin Uhlíř left us after spending 6 months in the ALOE team.

Martin did some great work for ALOE, we thank him a lot and wish him all the best for his time in Cordoba!