EMS Proceedings and Other Publications

Likn: A Flexible Platform for Information and Metadata Exchange




likn is a new web application which acts as a Content Management System (CMS) and discussion platform, but supplements these functions with an aggressive auto-hyperlinking architecture, and a powerful semantic metadata engine. likn draws on the success of collaborative editing as seen in projects like Wikipedia.org, and extends the model to include the democratic community development of relationships and metadata. The result is a unique platform for community-driven research, exchange and development.

Nodes in likn have associated descriptive text, which can either be locked, editable by a group, or freely editable. This text automatically hyperlinks to any and all matching nodes within the system. For example, if a node contained the text “...the Electroacoustic Music Studies Network, an international team which...” the phrase “Electroacoustic Music Studies Network” could potentially become a multilink-a menu which contained links to the nodes “Electroacoustic Music Studies Network,” “Music,” “Network,” “acoustic,” etc. Multilink generation is fully automatic. The frequency with which links are followed (and by whom) is stored in likn, and utilized to provide graphical navigational aids, recommendations, and to guess at association.

Users wishing to make more specific associations between nodes can utilize Qbot, likn’s semantic chatbot. For example, it’s possible to tell Qbot that “EMS06” is an instance of “conference,” which is a type of “event.” Events have start and end dates, and EMS06 would have inherited this trait, so it’s possible to tell Qbot further that “EMS06 begins on 23 October, 2006.” If you then asked Qbot what events were in progress on 23 October, 2006, it would return EMS06 as one of its answers. There are a few “prebaked” relationships (such as “instance of” and “type of,” used in the above examples), but through likn and Qbot, it’s trivial to generate novel relationships and properties. Thus, Qbot can be told that “EMS06 has the property ‘organised by’ with the value ‘Electroacoustic Music Studies Network.’”

likn also understands several relationships which create ontological constraints; for example, one can tell Qbot that “dark” is the opposite of “light,” or that “laptop computer” is the same as “notebook computer.” This capability opens up interesting possibilities for term-based translation, enabling the user to specify that “hopfhörer” is the same as “headphones” in German.

With so many assertions being made to Qbot, there must be some measure in place to verify the statements within the community. Every statement made to Qbot is reified internally and presented to the community at various points, allowing them to “agree” or “disagree.” When queried, Qbot will determine the popularity of the relevant associations, and qualify its answer in sophisticated ways if necessary. For example, if there are many votes on the statement “vinyl has the property sound quality with the value warm,” but the votes are fairly evenly split, Qbot may respond to a query with “it’s controversial, but many people believe vinyl’s sound quality is warm.”

Utilizing this data in many ways is one of the core concerns of likn, and to that end it has a flexible API, RSS feeds for every node, and the ability to generate RDF/XML vocabularies and OWL ontologies. Per-node RSS feeds eliminate the need for many features to be implemented explicitly in likn; users wishing to be informed of upcoming events can simply subscribe to the “event” RSS feed. Users who don’t want to hear about every event, just conferences, can subscribe to “conference.” In this way, each person in the community can determine his or her own level of involvement and notification.

The first implementation of likn was designed for and on criticalartware, which is a group/application/platform designed to address the histories of new media, software-as-art/art-as-software, and connections between the current early stage of artware and the early moments of video art. The power and flexibility likn was able to give to the criticalartware community was profound, and enabled a uniquely rich dialogue to begin. An open source release of likn is planned once the code is further production tested, but in the meantime, criticalartware is anxious to share our experience with our most unusual Semantic Web application, likn.