We believe that it is possible to build groupware services that are open, interoperable and respectful of user rights. Our approach utilises FOAF and RDF to enable decentralised data storage across the network in preference to storing it centrally at a single service. A core tenet of our thinking is that no private information is held about the service users. This approach solves a number of existing problems but introduces new complexity in the areas of identity, authority and data freshness.
The established online groupware services exhibit a number of shortcomings which make them at best inconvenient and at worst dangerous for use by many organisations and individuals. Invariably they:
We are investigating how decentralization of user data can eliminate these shortcomings. In our model we expect users to publish descriptions of themselves and their group preferences utilising FOAF and related vocabularies. By adopting the RDF model for data interchange we expect to achieve reliable interoperability with other RDF aware tools. Any tool capable of emitting the appropriate triples can be used to publish the required information. Storage of that information is decoupled from the service and so can be reused with many different services.
Authority for making assertions about users can be confirmed through the use of one-time identification tokens sent to the email address of the user. These tokens can be used to confirm authority for that user allowing them to maintain the list of authoratitive description URIs that should be used.
We have observed the difficulty caused by vandals in the case of open Wikis, and spammers in the case of open comment systems on weblogs. Therefore control over groups is desireable and we are examining different models ranging the spectrum of democratic to despotic. In particular, we want to allow the community to determine the membership policy - whether completely open or invite only. It may also be possible to allow free membership but filter each member's view of the group by their extended social network. It is imperative that that control mechanisms do not hinder easy initiation of communities.
Many tools and facilities such as contact lists, shared calendars, group weblogs and rss feeds can be built through aggregation of self-published information which is only kept for display purposes. Aggregation can take many forms - for example news items, book lists, or bone china auction events, or wants and needs in the case of a mini marketplace.
A few facilities such as email reflectors need no central storage of data. Some facilities such as wikis are not conducive to distributed storage but this can be alleviated by providing full read access to the underlying data files.
Not every group wants or needs to use the same collaboration mechanisms. A software project might want email distribution and bug tracking, whereas an event organiser would want a shared calendar and notification service. Our goal is to be non-prescriptive about how services are used and to enable different modes, mechanisms and forms of collaboration.
Semantic Planet is developing FriendSpace (http://www.friendspace.com/), a decentralized, interoperable environment for group collaboration. FriendSpace is designed from the ground up as an application of FOAF, and a FOAF document is the lowest denominator needed to bootstrap a group. FOAF provides the core properties upon which FriendSpace will contruct groups. The foaf:interest property can be used to identify when to create groups and who should be a member and in conjunction with foaf:knows it can assist in identifying other people who may be stating the same interests. FriendSpace may also use foaf:knows to determine additionally membership elegibility into the group, or to filter the view of the group's activities that a member may see. Properties such as foaf:weblog can be used to aggregate published content for the community.
FriendSpace can help FOAF by providing a compelling application that will encourage users to create, update and use FOAF descriptions about themselves. For FOAF virgins, FriendSpace will be able to generate a very simple FOAF document for users without an existing document, simply based on their email address and stated interest.
FriendSpace is pre-alpha vapourware in the traditional semantic web sense. If you want to participate in an alpha trial of the service, add
<foaf:interest rdf:resource="http://www.friendspace.com/groups/galway-foaf-2004"/>
in your FOAF and submit your FOAF url at http://www.friendspace.com/join
Semantic Planet (http://www.semanticplanet.com) develops software and services based on semantic foundations. It was formed in 2003 by Ian Davis and James Carlyle, who both plan to attend the workshop.