LSDIS > Projects > SemGrid

SemGrid: Semantic Discovery on Adaptive Services Grid

Support for semantics requires the ability to extract semantic metadata and create semantic associations, process large amount of metadata leading to better search, integration and analysis. SemGrid project is a collaboration between two complementary research efforts funded by EC and NSF, respectively: the Adaptive Services Grid project (ASG) funded by FP6-IST addresses composition, adoption, and distribution aspects of semantically described services, while the Semantic Discovery (SemDis) project funded by NSF-ITR addresses the synergistic semantic and analytical aspects. Two specific areas of proposed collaboration are:

  1.  Using Semantic Associations in WS-Discovery and Dynamic Web Process Composition and
  2.  Computing Semantic Associations over the grid.

Discovery and Composition of Semantic Grid Services

The UDDI protocol is the de facto standard for web service publication and discovery. Yet, it has poor performance and lacks the ability to use semantics. We aim to create a discovery and publication application, Semantic Web Services Registry building on knowledge from METEOR-S and SemDis that uses semantics by making it ontologically aware. Published semantically annotated web services will be indexed, in an extended version of BRAHMS, using their data and functional semantics. This allows us to protect a document's semantic identity while allowing faster discovery using their semantic annotations.

We will create a way by which the computer will be able to take annotated web services from the Semantic Web Services Registry and compose them into a web process that will perform a user defined task for which no single web service is registered. Much of work done by this process will be extension of the SemDis discovery and ranking of semantic associations. We envision accomplishing this by having the goal for a web service described as functional ontology class. With the associations between this describing class and the rest of the ontology, we will create a skeleton workflow of the task. The ρ path operator will be used to discover paths and the ρ Iso operator to discover alternate paths.

Semantic Association Discovery on Grid

Grids are a way to provide supercomputing power to the average person. The use of Grids in discovery of semantic web services is important for two reasons. First, the web is a distributed environments itself, as such it follows a model similar to the web will be most effective. The computation of discovery of a few million nodes and relationships may be feasibly done by one server class machine, but when the computation is done at the scale of web, we are talking about a few billion nodes and relationships, and as number nodes increases computation over greater path lengths will occur which leads to an exponential increase in the number of paths. Therefore, it makes sense to try to distribute the data and computation of this information and Grid based computing is the most plausible way to do this. We will be using the technology developed with the ASG group to develop the necessary Grid algorithms and computation strategies.

"SemGrid: Semantic Discovery on Adaptive Services Grid," is funded by NSF Award # IIS 0545243, October 15, 2005 - December 31, 2007. Principal Investigator, PI: Amit Sheth, Co-PIs: I.B. Arpinar, K. Kochut, J.A. Miller. This project involves collaboration between LSDIS's SemDis and METEOR-S projects, and the European Commission sponsored Adaptive Service Grid project. UGA Contact: Amit Sheth. ASG Contact: Mathias Weske.