The present Semantic Web uses sychronized communication with different platforms (each with its own API) to allow clients to add, update or remove knowledge with server-side reasoning. This will prove not to be
feasible for realizing complex Semantic Web applications where m clients and n platforms need m*n connections, have to know what platforms they will communicate with in advance and have no guarantees that platforms are
persistently accessible nor that the present knowledge is available when the query is made.
The Semantic Web needs to profit from the same advantages as already seen on the Web itself: the publication of data, a simple means to act on that data (like HTTP) and the
distribution of data across servers without adding complexity in the client. A middleware should not only provide for asyncronous and standardized communication but also, in a Semantic Web
that is supposed to enable critical tasks through knowledge sharing, mediated and guaranteed coordination.
We identify the work on Linda and tuplespaces as highly relevant to meet this need. Data is published into a tuplespace, and acted upon using the simple primitives of the Linda language. Furthermore communication in this paradigm is loosely coupled: the client need not concern itself about the location of the data which could be distributed anywhere on the network, nor when the data was published, nor who published the data.
We extend the Linda coordination language and the tuplespaces paradigm that it uses to enable the exchange of knowledge, rather than data, in a
loosely coupled fashion between Semantic Web applications. Semantic Web Spaces realizes:
Check the latest news from our research, read more about the project, access our publications or contact us for more information!