Feb 27 2009
Happy Birthday KiWi!
Congratulations! Tomorrow, the KiWi project is turning one year! Time to look back and see what we have achieved, and I dare say it is quite much, partly more than we expected: we have finalised the first Open Source (pre)release of the KiWi core system, we have learned a lot about the properties and challenges shared by all semantic social software in the areas of semantic tagging, reasoning and information extraction, we have developed a much clearer vision in the use cases, and we have been quite good in presence on conferences and other dissemination activities. Read on to learn more …
KiWi Core System. A major milestone of the first project year is the first Open Source (pre-)release of the KiWi core system. The KiWi core system is a flexible platform for building different kinds of semantic social software applications on top (currently the Semantic Wiki and the TagIT application). It provides all the core services required in such applications, like editing and tagging, the storage of content and associated meta-data, its own triple store, transactions and versioning over content and meta-data, linked open data server, and many small features semantic social software developers will like (like convenience services for working with ontologies or SKOS thesauruses, etc.). And finally: the KiWi system also has a quite neat user interface by now, rarely found in Semantic Web applications.
Technology-wise, the KiWi system is a component based architecture building on JBoss Seam and Java EE. The main principle is that everything is a content item, and that a content item always consists of human-readable content that is connected with RDF relations and may be annotated by users with semantic tags. This mixture allows content to be shared between applications: a content item that is created in the Wiki application can at the same time be displayed in TagIT as a location on the map or (later) as a blog post in the Blog application. This is made possible by the flexible structures offered by RDF. A particularly salient aspect of the KiWi core system is that it makes using RDF much more convenient for Java developers (no more manual triple store queries) by offering a flexible facading through interfaces annotated with special Java annotations that map Java methods to RDF properties.
The KiWi core system is of course not finished yet: we are planning to extend it with so-called “knowledge spaces” (separating content into different logical units), permission management is also still open, we’ll include social networking capabilities, we’ll integrate Sun’s Community Equity, and we’ll build additional applications on top of KiWi, including the personal “Dashboard”, a “Blog”, and a mobile application to access some of the functionality even when away from the computer. And then we are of course looking forward to the integration of the enabling technologies in the system!
Of course, a KiWi that is only one year old is still to young to really fly, and it will even still stumble from time to time, so expect bugs. But nonetheless, we are now inviting all of you, particularly (Open Source) developers and early adopters to try out and work on the KiWi system, build your own applications on top of it, or just experiment and give us feedback. You can download the latest KiWi release at the Kenai Download Site or just visit the Showcase (updated regularly, may crash from time to time).
Enabling Technologies. Working the “Wiki Way” provides a lot of challenges to research disciplines that are traditionally based on very formal means of knowledge representation, like KiWis “enabling technologies” in the areas of reasoning, information extraction, and personalisation. Established approaches – e.g. for reasoning – make assumptions about the data like consistency or pre-defined knowledge models that are not acceptable in wiki environments. Usability is also a major issue – one can hardly expect users to search in KiWi using SPARQL or similar languages.
Enabling Technologies in KiWi have therefore shifted from classical knowledge representation (like in RDF) to working with more Web 2.oish means like tagging. Tagging in KiWi is of course semantic: a content item can be tagged with another content item, thus immediately embedding the tagging into a (semantic) context.
The components developed in the enabling technologies will all build on this semantic tagging: the group in Munich is currently developing a query and reasoning language that allows to work on content items and their taggings using so-called label-keyword queries, the group in Brno is developing tag recommendation based on latent semantic analysis which will also support disambiguation, and the group in Aalborg has been working on personalised recommendations based on the tags given to content items and the tags most commonly used by the user.
Of course, tags are also represented as RDF triples, making it possible to also work on the tagging data with more traditional means (e.g. SPARQL, rule languages developed for RDF, or Linked Open Data).
Use Cases. Work on the use cases throughout the first year has been characterised with identifying what is needed by our partners and figuring out what a research project can actually deliver. In the first six months, we identified requirements. A common requirement in both use cases is the integration with existing systems, and a major step into this direction has already been made by KiWi by providing all data as Linked Open Data. Also, the Sun use case (software knowledge management) already heavily influenced the directions the core system is going: things like semantic tagging, the new user interface, the planned social networking component and dashboard, as well as the foreseen integration of Community Equity witness this.
The Logica use case (project knowledge management) on the other hand demonstrates the challenges faced by semantic social software when trying to deploy it in a very formal environment like project management. We are still discussing on how the KiWi system could mix Wiki-style with form-based working, and how KiWi could connect to ERP tools like Navision.
Dissemination. Even throughout the first year, the KiWi project managed to achieve a quite high visibility in certain communities. We have developed a very detailed dissemination plan that describes the overall dissemination strategy and the activites that we aim to take.
On the scientific side, KiWi co-organised several well-attended events like the SemWiki08 workshop at ESWC’08, the Social Web and Knowledge Management workshop at WWW’08, the I-Semantics conference, and the ReasoningWeb Summer School. Also, we have been present with presentations, posters, and tutorials at ESWC’08, SofSem’08, DLAC III, XInnovations, and many more. Also, we are co-organising the Ontolog series on Semantic Wikis, which established the Semantic Wiki topic as one of the mainstram topics in the Semantic Web community.
On the professional side, we have given early demonstrations to industry users (e.g. at the IFRA Conference on the Future of News Publishing), managed to build several additional projects on top of the KiWi platform (TagIT is a result of one of them), and there are still more to come. Also, we are receiving requests about KiWi on a weekly basis even though no software has been available until today.
And last but not least, we have a quite neat project identity with a nice and informative website, public relations material, the Planet KiWi blog, and the KiWi development community platform and many small goodies that make the project visible.
So again: Happy Birthday KiWi! We are looking forward to your next year! And thanks to all of you for making it a success!
No responses yet
