March is time for KiWi annual meetings. This year’s meeting took place from 10th-12th March in Mattsee close to Salzburg, again at a very nice location. The Annual Meeting was focussing primarily on the use cases (which are supposed to start with evaluation now) and the dissemination and exploitation activities (which now go into the last phase). Here is a short summary of the meeting:
Core KiWi System
As usual, the annual meeting began with a presentation of the current state of the core KiWi system. Rolf presented the novel implementation of Semantic Forms (based on RDFa) in KiWi. KiWi’s Semantic Forms can be used as alternative editors for a content item and directly update the RDF metadata inside the KiWi system. Planned improvements of Semantic Forms are the support of RDFa object properties and the automatic generation of Semantic Forms out of types specified in an ontology.
Thomas then briefly presented the latest version of the TagIT implementation on top of KiWi. TagIT is accessible at http://tagit2.salzburgresearch.at and will go into a productive use by end of March (accessible under http://tagit2.salzburgresearch.at). Arpad concluded the Salzburg Reseach part with a presentation of the new vocabulary management tool for KiWi, which is at the same time a first experiment for using GWT (Google Web Toolkit) in conjunction with KiWi.
Peter Dolog summarised the various personalisation features that have been implemented in KiWi. Personalisation is available at several levels in KiWi: it can recommend related content to the currently displayed page based on the user’s preferences, it can recommend content interesting to the user based on his activities in the system, and it can be used for personalising the ranking of search results in KiWis search interface (so-called personalised search). In all cases, it is a prerequisite that there is sufficient data for the personalisation from previous (tagging-) activities of users.
Klara then presented the latest progress in using the query language KWQL in KiWi. Beyond ordinary search, KWQL is capable of issuing structured queries to the documents contained in the KiWi system. KWQL is available as part of the core KiWi system for some time already and can be used as an alternative to the default search mechanism.
Jakub showed the new explanation service for the reasoner in the KiWi system. The explanation service allows developers to inspect the justifications why the reasoner has inferred a triple in a graphical manner. It can also be used to show explanations to the user for certain kinds of behaviour (currently inferred types and incoming/outgoing relations).
Marek concluded the presentation of the KiWi core system with a demonstration of the various information extraction functionalities in the KiWi system. Currently, information extraction is used for tag recommendation. He is currently also working on using it for automatically recommending RDFa annotations for the textual content of a document.
Sun Use Case (Software Knowledge Management)
KiWi’s Sun Use Case is concerned with knowledge management in the SunSpace intranet of Sun Microsystems, now part of Oracle. Josef Holy and Peter Reiser started with presenting the three storylines they aim to evaluate as part of the use case: concept model management (i.e. how to manage the company’s internal thesaurus using KiWi and PoolParty), Text Extraction and Tag Recommendation (both, based on the company thesaurus and using free tags), and Searching and Browsing through the SunSpace intranet using KiWis Semantic Search functionality.
Sebastian and Mihai continued with a presentation of a reimplementation of the Community Equity system in KiWi. The new implementation uses a much simplified implementation as compared to the currently used system and gives at the same time more flexibility to admins and developers. Still unclear is how well the system scales in the presence of huge activity logs.
Further steps concerning the implementation and evaluation of the Sun use case are the connection of Confluence with KiWi using a plugin to “augment” Confluence with KiWi’s semantic technologies, the connection of KiWi and PoolParty, as well as the migration of content, particularly activity logs, from SunSpace into the KiWi system to preseed the system for evaluation.
Logica Use Case (Project Knowledge Management)
Daniel Grohlin, Karsten Jahn and Peter Axel Nielsen presented the implementation and scenarios of the Logica Use Case, which is concerned with managing project knowledge. The evaluation will cover the scenarios “project planning”, “project monitoring”, “development or project work”, and “process design”. All four scenarios will be evaluated using a combination of the KiWi system and the proprietary Logica application that complements and accesses KiWi using web services. The use case will be evaluated using an “agile evaluation” that refines the tests and implementation in several evaluation cycles (probably three). Particularly noteworthy is the feature matrix developed as part of the test plan that will be extended by the other partners and used for the Sun Use Case as well as as input for the dissemination and exploitation activities. The feature matrix is also useful to verify to which extent the KiWi technologies are covered by which of the scenarios.
Dissemination and Exploitation
The last of the very important topics of the KiWi Annual Meeting was the discussion on dissemination and exploitation. While dissemination was quite successful in the first two years running with only little resources, the project now moves on into the third phase where it is important to increase the resources spent on making the project known and “selling” its results to the outside world. To this aim, John started with presenting a timeline for the next year, highlighting the most important milestones of the dissemination and exploitation activities. As a major step, KiWi will relaunch its websites and split them into a “research project page” targeted primarily at researchers (as it is now), a “system page” targeted primarily at decision makers as well as a “community page” targeted at developers that want to make use of the KiWi technologies. Further steps in the dissemination area are still confidential, but stay tuned on updates! And don’t miss the KiWi release party for version 1.0 – it will probably take place in September!
