Corrib Cluster Blog

I would like to invite everyone to take part in the evaluation of the semantic and social technologies for digital libraries. The evaluation benchmarks search and browsing solutions delivered in our semantic digital library called JeromeDL [http://www.jeromedl.org/] against standard services offered by one of most popular open-source libraries - DSpace.

Please feel free to enter http://q.digime.name/ and help us with evaluating our prototype solutions.

Researcher - Software Developer

(please note that the recruitment process is now closed)

The Digital Enterprise Research Institute (DERI) is the largest semantic research organisation in the world. DERI offers a stimulating, dynamic, multi-cultural research environment with excellent ties to research groups worldwide. This is a unique opportunity to join the effort of bringing research prototypes to industry ready within DERI, in collaboration with our research and industrial partners will play a key role in making next-generation semantic computing systems a reality. DERI offers a unique opportunity to develop one’s career in the world-wide renown and industry strong research environment.
 
The Person
• Ability and willingness to work in a international team based environment developing state of the art software solutions on time and to specification
• Motivated and proactive attitude to take ownership and initiative in all work assignments
• Excellent analysis and problem solving skills
• Strong design, development & testing skills
• Excellent communication skills, verbal and written
• Excellent command of English, both verbal and written
• Ability to tackle wide and varied tasks
• Creative Thinking

Essential Skills
• Solid industry experience using many of the following:
• Very strong core Java
• Web based UI: JSP/Servlets/Applets/JavaScript/AJAX
• Good expertise with automated testing frameworks such as JUnit
• Good knowledge and experience with Semantic technologies
• Good knowledge of object-oriented design principles and design patterns with an understanding of their application within Java
 
Desirable Experience & Background (inc. qualifications):
• Knowledge/Experience with distributed systems and service-oriented design principles
• Knowledge of user interface design principles
• Experience/Knowledge of document processing and search techniques
• Experience/Knowledge of XML processing and related technologies
• A relevant post graduate degree (MSc) or relevant industrial experience

The position is full-time, located at DERI Galway. The duration of the post will be for 9 months in the first instance. The salary is commensurable with qualifications and experience. An early start date is preferable as the position is now open. A panel for future similar positions may be formed.
Informal enquiries about these positions may be made to:
Sebastian Ryszard Kruk, Researcher and Project Manager, Tel +353-91-495213 sebastian.kruk@deri.org

Application procedure: Candidates are requested to submit a covering letter, CV (Word or PDF format only) and the names and addresses of at least three and not more than five referees via e-mail to;
hr.ie@deri.org

The contest is finally over.

While the dust of shattered tags still floats in the air, and piles of dead mice lay along curbs, we watch the mouseslingers stroll down the alley to the Café, we greet the one that remained - The Fastest Clicker West of Dublin.

Two weeks ago, at noon - almost 50 of them took the challenge; they wanted to show off to the world their keen eyes, quick draws, and precise clicks. The task was tough - shoot through 216 tags hiding in large flocks in the clouds and behind the trees. There was no mercy. Each and every of selected tags had to be clicked. But they made it. And today was the day of their victory. The price was worth the trouble - the became the fastest clickers west of Dublin. Three of the fastest of them were awarded with distinctions.

• Mateusz “kaczor” K - The Unbeatable Tags Slaughterer (13.5 tags / minute)
• Michał “hoook” N. - The Highly Lethal Mouse-Slinger (13.5 tags / minute)
• Mateusz “spajk” M. - The Advanced Mouse Driver (12.5 tags / minute)

Today they have been awarded.


Watch them,

envy them,

thinking

… damn, this could’ve been me ...

As you might have noticed, spajk was already riding towards the sunset. But he will soon be back to receive his award.

On behalf of Maciej, Ewelina, Stefan, and myself, I would like to thank everyone who participated in this hard evaluation process. You all were great and that is why we decided to award all of you, to thank you for your help. We’ve gathered a lot of interesting and valuable data. We will keep you posted on our publication process, as each of this articles will be a tribute to you.

Here’s to you …

Cast (in alphabetical order)

• Cosmin Basca
• Uldis Bojars
• Smitashree Choudhury
• Stéphane Corlosquet
• Maciej Dąbrowski
• Brian Davis
• Prof. Dr. Stefan Decker
• Jakub Demczuk
• Siobhán Dervan
• Andrew Gallagher
• Ioana Giurgiu
• Jędrzej Gontarczyk
• Adam Gzella
• Iza Gzella
• Armin Haller
• Dr. Siegfried Handschuch
• Andreas Harth
• Dr. Conor Hayes
• Aidan Hogan
• Jacek Jankowski
• Mateusz Kaczmarek
• Hak Lae Kim
• Sheila Kinsella
• Ewelina Kruk
• Sebastian Ryszard Kruk
• Hsiu-Fen Grace Lin
• Nikos Loutas
• Mateusz Marmołowski
• Fergal Monaghan
• Tadhg Nagle
• Dr. Ke Ning
• Vit Novacek
• Michał Nowacki
• Ina O’Murchu
• Szymon Pająk
• Łukasz Porwoł
• Maciej Salamonik
• Krystian Samp
• Magda Samp
• Brahmananda Sapkota
• Dmitry Shingarev
• Oana-Elena Ureche
• Dr. Tomas Vitvar
• Dr. Brian Wall
• Maciej Więckowski
• Steve Williams
• Maciej Zaremba
• Wojtek Zaremba

This weekend I took a couple of minutes to get away from working on the CHI2008 paper, and improve our version control infrastructure for Corrib.org. Until yesterday, we used cvs.corrib.org to keep maintain source code of our publications and any projects that were inappropriate to be published on http://sf.net .

Here is what has changed:
• complete archive of cvs.corrib.org has been moved to be handled by Subversion, which I hope will improve our performance, at http://svn.corrib.org/
• ViewVC has been installed to allow browsing our repository across different versions; go to http://vc.corrib.org/ to access this function
• Finally, a RSS feed with recent changes (commits) done in our version control repository is available at http://vc.corrib.org/rss.xml

I hope every one will enjoy this improvements; you will need a labs.corrib.org credentials to access these features, though.

– Sebastian

I am very pleased to announce the FOAFRealm 2.0.
The installation package should are available at the project web site at sourceforge.net.



Release changes
• You can create an invited only network (Orkut style).

• Improved web-browser compatibility issues.

• In contrast to the beta version, this release works with IE5++ and Opera 9++

• Added simple foafbrowser

• The code of the simple captcha is slightly rewritten, so now it can work on linux servers.

• 3 to 5 times shorter loading time because dojo is removed.

Minor problems

• Problems with the SMTP server
• The none picture now work

• Improved DIVs at the login page

• New layout of the buttons (now dojo is only for validation purposes)

• Added refresh captcha (still not perfect)

• After save you have to wait 10 seconds for another try

• The friend profile is updated after a change

Developers info
• The main part of the db module is rewritten.

• No more dependent on a tomcat/lib/(sesame jars)

• Now it uses own sesame jars and thus to make it work, 

• Most likely you have to remove all sesame jars from tomcat/lib

Installation notes

1. Download the installation package.

2. Extract the war files: dfoaf-ui.war, dfoaf.war

2.1 Put those files to your tomcat/webapps folder

2.2 Download foafrealm.jar and foafrealm-toolbox.jar and put them to your tomcat/lib folder

3. You need also an RDF repository. The system was tested with sesame 1.2.5. If you decide to use sesame, you also need to put sesame.war to your tomcat/webapps

4. Run tomcat the wars will be automatically extracted.

5. The webapps need some customization.

5.1 You have to set up the repository. 

e.g you can add the following entry to your sesame/WEB-INF/system.conf file:




<repository id="foafrealm-repository">
     <title>Native RDF repository</title>

     <sailstack>

          <sail class="org.openrdf.sesame.sailimpl.sync.SyncRdfRepository"/>

          <sail class="org.openrdf.sesame.sailimpl.nativerdf.NativeRdfRepository">

                 <!-- The directory for storing the native SAIL's data files. -->

                 <param name="dir" value="/tmp/nativerdf/"/>

                 <!-- The triple indexes that should be used -->

                 <param name="triple-indexes" value="spo,ops"/>

          </sail>

     </sailstack>

     <acl worldReadable="true" worldWritable="true">

              <user login="testuser" readAccess="true" writeAccess="true"/>

     </acl>

</repository>

5.2 Now you can change to default settings, if you use a different port than 8080 and/or address than localhost the soa service address is in foafrealm.jar which is in your tomcat/lib foafrealm.db, dfoaf-ui/WEB-INF/classes/foafrealm.properties - dfoaf-ui 
 
You have to have a look at the application parameters and set them to correct values.

storageDir=/tmp/dwn2/

inviteFriends=true

inviteOnly=false

SOAaddress=http://localhost:8080/dfoaf/soa/
confirmRegistration=true
localShowPictureAddress=http://localhost:8080/dfoaf-ui/ShowPicture
registerCredentials=cmVnaXN0ZXJAZm9hZnJlYWxtLm9yZzpjaGFuZ2VpdCEh

• Create the storage dir at the path that storageDir variable points to.

• Make sure that the ShowPicture servlet address is correct.

• The SOAaddress should point to the dfoaf.war extraction point.

• The other parameters allows you to control the registration process.



5.3 SMTP configuration.
To make confirm registration and forgot password work you have to edit the SMTP parameters. The parameters are in the mail.properties file located in the foafrealm-toolbox.jar


6. Now you can visit http://localhost:8080/dfoaf-ui/

7. Using FOAFRealm in your own tomcat application is pretty easy.
You can just extend the dfoaf-ui project, or copy-it to your web-application.

You should use it by calling jsp-tags:

-profile - to access the user’s profile e.g.


<%@ page contentType="text/html; charset=UTF-8"%>

<%@ taglib prefix="foafrealm" tagdir="/WEB-INF/tags"%> 

<foafrealm:profile />

-logout - to expose the logout button


<%@ page contentType="text/html; charset=UTF-8"%>

<%@ taglib prefix="foafrealm" tagdir="/WEB-INF/tags"%> 

<foafrealm:logout redirect="index.jsp"/>

-tomcat - to perform the tomcat authentication e.g.


<%@ page contentType="text/html; charset=UTF-8"%>

<%@ taglib prefix="foafrealm" tagdir="/WEB-INF/tags"%>

<link href="${pageContext.request.contextPath}/css/foafclient.css" type="text/css" rel="stylesheet" />

<foafrealm:tomcat />


You need to edit your web-application context in the tomcat/conf/server.xml file
and add realm and valve, for example:
 


        .....

          <Context className="org.apache.catalina.core.StandardContext" cachingAllowed="true"
              docBase="myapp" path="/myapp" 

              privileged="true" reloadable="true" > 

            <Realm className="org.foafrealm.realm.FOAFRealm" debug="99"/> 

            <Valve className="org.foafrealm.realm.FOAFValve" debug="99"/>

          </Context> 

        .....

      </Host>

    </Engine>

  </Service>

</Server>

8. Enjoy,

I would like to thank all the people who gave me their feedback, especially Adam (SOA, comments on UI), Władek (RDF2Go), Peyman, and the everyone from Corrib and FOAFRealm groups.

Sławomir and Sebastian

JeromeDL 2.1 is almost released. I am fighting with MultiBeeBrowse+Exibit+TagsTreeMaps integration; a lot to be done, and I am the last man standing.

I decided to reveal some of my plans related to new features and releases of JeromeDL.
Once version 2.1 is out, you will be, finally, able to browse the content of digital library using various, quite sophisticated, solutions:

• Exibit, the award winning browsing solution from SIMILE, will allow to quickly filter out desired resources, based on a number of properties of resources; if your books are annotated with spacial or time information, special views: Timeline and Google maps, will show you where and when.
• MultiBeeBrowse, homemade collaborative faceted browsing component, will give you a lot of freedom to browse from the place where you were to sets of resources you would not find any other way. Unique features like non-linear history of browsing, sharing browsing experience with your friends, and recalling your previous browsing sessions, might be a little mind-boggling; I am certain you will find them useful, though.
• TagsTreeMaps, finally, since people seems to care more about tags than any structure information (so far), another homemade solution will give you a unique experience of tags-based filtering. And you will see not only resources in the library, but also your bookmarks in SSCF, and … your posts in del.icio.us

You can already preview MultiBeeBrowse and TagsTreeMaps in our online, informal knowledge sharing portal: http://notitio.us.

Next versions of JeromeDL, should also feature following components, currently developed by the group projects ran together with Gdansk University of Technology:

• Semantic Query Expansion engine, based on extrapolated user profile, and other context information, will finally make its way to JeromeDL. Its first implementation in Elvis-DL, prececessor of JeromeDL, became absolute once we made a high-sky jump to current JeromeDL 2.0. I cannot wait to see, the “tell me why button” finally in action; with all the data we can now handle in JeromeDL, it is bound to be a success.
• MARC21 and OAI - I hope that we will finally close this chapter of unsuccessful attempts to support these standards. MARC21 was supported by Elvis-DL, but somehow did not make it fully to JeromeDL. OAI-PMH, although said to be implemented by Bugi, never become mature enough
• collaborative photo annotations - well, we can handle photos in JeromeDL, even now; but there is a lot of features I always missed, including photo-specific type handling. If you know ESP game, you should enjoy the asynchronous version of it; I know I will.
• augmenting video streams with community annotations - if you ever watched a movie, especially a tutorial video, you might have drifted away the same way I do, when there is nothing happening except from one man telling the story. And I guess, often you could find yourself saying: I know more about it, or: I don’t get a single bit. The new feature of JeromeDL will allow, first of all, a community of users to bind their comments (blogging support in JeromeDL) to the particular part of a video stream, using time tags; once the video is augmented with more interesting information notification will be displayed once playing it back, so that users can pause it, and read more, follow some interesting links, etc. I just wonder, will we make it on time, or will YouTube’s gonna be first ?

Once I am through with the 2.1 release, I hope to wrap up all my research on temporal ontologies, and eventually put it into practice with versioning support in JeromeDL. During recent tutorial on Semantic Digital Libraries, there were people asking me if they can keep the whole history of changes made to the resources in JeromeDL; they can’t, so far, but I hope it will be possible in a near future. Will it transform JeromeDL into a wiki-model application? Probably, as soon as we will figure out what is missing in this picture.

During ESWC2006, I managed to have a short chat with Tudor Groza, who was presenting SALT project (Semantically Annotated LaTeX); it was a follow up, a rather late one, on our previous discussion on how to deliver support for SALT in JeromeDL. The first time I thought about this integration, I jumped to conclusions that we will need to build a special viewer, which would allow JeromeDL to present users with the exact part of the document they have filtered out, e.g., with MultiBeeBrowse; an alternative would be to use JavaScript support in Adobe Acrobat, but I still do not have a clue how to invoke it while downloading the PDF file from JeromeDL.

This time when I talk to Tudor, he presented me with a publishing component, that could, e.g., take couple of PDFs with RDF information in them, and generate a concise representation of a whole collection, presenting most important information from each of publications. This has triggered a new way of seeing how SALT could actually be used in JeromeDL.
During our tutorial at WWW2007, I have realized that what is missing in JeromeDL, and what I always wanted to have, is ability to use the fine granularity of documents (chapters, pages, media parts, attachments) and the documents themselves, and combine them into new resources - collections. At the moment, user can only have 4 types of pre-defined collections: by bibliographic type, by author, by keyword, and by category. What we should be able to do is to create a resource that:

• consists of other resources, or their parts, or both;
• is a collection of resources being a set of results of given query;

Now, when we will put SALT in this picture, we can see that we could actually get even finer granularity of publications, and use this small parts of resources to create new ones.

Another idea, which pop-uped was triggered by the adaptable presentation boxes done by Filip in his Master Thesis; he extended MultiBeeBrowse with much nicer presentation style. In my opinion, having all this cool annotations inside of documents, we could present each resource with much more information than we do at the moment, while controlling what is shown, and what is not, with this adaptable boxes. Further more, if we use IKAR, another component created by Filip, we could present information about resources referenced from the one being currently viewed; and the adaptive hypermedia solutions would do the trick to make sure it is presented how users expect it to see.

Finally, the DRM feature, which I have been asked about several times; so far, I know what needs to be done, and even current FOAFRealm can help us with the trick. I am out of resources at the moment; hence, it is very hard to predict when this feature will be delivered. Hopefully, pretty soon.

As you can see, there is a lot of going on in JeromeDL; and I hope we are going the right direction. One more news I want to share with you, is about the status of the Enterprise JeromeDL. It turned out that Mark’s team wants to write JeromeDL from the scratch in one of scripting languages. I still cannot decide whether I like it or not. One thing is for certain, we will not see the industry stable version of JeromeDL with all current feature in the near future.

Technorati Tags:
jeromedl, release plans, features

It struck me that so much research is done in the semantic web community, but many of the people can either “spell” metadata or semantics; and very often people from both semantic and non-semantic community cannot tell those two (metadata and semantics) apart.

Technorati Tags: digital libraries, metadata, research, semantics

(more…)

Today we met Tom Gruber; for those that do not know, he is one of the key people in the Semantic Web research. To be frank, he came looking for me on Monday; which caught me really surprised.

Tom was very interested in hearing about our research on semantic digital libraries. We started with short introduction to JeromeDL, and quickly get to the matter of bibliographic ontology. We talked about MarcOnt ontology, our research, upcoming new release, and FRBR ontology created by Ian Davis. The open question remained - who else works on a bibliographic ontology? The only  other projects I could refer were the RDA and Dublin Core collaboration and DCMI Abstract Model and RDF.

Tom mentioned about TagCommons, and that they will be putting a service/metadata online with tags coming from various sources - cross application (tag watcher). He mentioned that searching on this information will be a challenge worth taking.

I introduced him to notitio.us, including SSCF, TagsTreeMaps, IKHarvester and MultiBeeBrowse components deployed there. He was very impressed with SSCF and TTM, and said that we should make sure TTM will hook to the tag watcher service.

We decided to continue collaborating on the research related to tags and semantics in the TagCommons. I will see how to adapt TTM to his source of tags.

technorati tags:semtech, tomgruber, tags, tagcommons, notitio.us, tagstreemaps

Blogged with Flock

Today I presented JeromeDL, and other Corrib components from eLITE project, to the Charles Petrie’s group at DERI Stanford. The talk “Digital Libraries of the Future -  Use of Semantic Web and Social Bookmarking to support E-learning in Digital Libraries”  went quite well, with many interesting, and sometimes intriguing, questions; even thought I hardly managed to make it short

All the comments I got were really encouraging, and it looks like the audience did like this talk. I have also got an interesting comment to follow up on my Real-world Semantic Museum scenario.

After the talk I briefly met Charles to discuss how Digital DERI and JeromeDL at al. can come together. The we had before was to use JeromeDL as a publishing platform within DD. However, as Charles pointed out, we should find a real use case which can be resolved by JeromeDL. An opportunity for working together appeared with SSCF project. Here are some ideas:

• Digital DERI could make use of categories created and maintained by community of users of SSCF component (in one of JeromeDL instances or, e.g., http://notitio.us/ service). These categories would be less rigid by would answer the needs of the community. To deliver this feature we would need to address following problems:

• each bookmarks category have a unique URI (it does at the moment) which after clicking can show this category, even though its author is not our direct friend - we can introduce a notion of “temporary friend”, and allow users to “add” this author as a friend from SSCF interface
• there should be an RDF information delivered behind each bookmarks category - I think eRDF, RDF/A and/or GRDDL technologies will not be over estimated in this case
• we are able to express SSCF information in SIOC, but as you know SIOC does not handle hierarchical information, like categories, in which case we should do SKOS export

• we would need a mediation mechanism - should be pretty easy with both speaking RDF
• we would need a notification mechanism - can OpenID help us with that ?
• we need to address the problem that DD has a completely open access to profile data, so that everyone can add any information about anyone.

technorati tags:presentation, meeting, jeromedl, sscf, deri, stanford, digitalderi, cooperation

Blogged with Flock