Automatic Composition of Dataflows using Semantic Graph-Transformation Models of Components

A poster presentation written by Anton Riabov, Anand Ranganathan and Zhen Liu. It was presented at the ESWC2007. It is about Automatic Composition, Graph Transformation, Semantic Modeling of Components, Planning and Workflows

[edit] Abstract

In this paper, we consider the problem of composing a dataflow graph in an automated manner from a number of reusable components. A dataflow graph describes the flow of data between different components as a DAG (Directed Acyclic Graph) and is composed in response to a user query for certain kinds of data. We propose a novel way of modeling components using semantic graph transformations. Each component is associated with a semantic annotation that describes its inputs and outputs using RDF graph patterns. A user query is also represented using RDF graph patterns. In this paper, we provide a formal description of the component model and define the conditions under which components can be connected in a dataflow. In addition, we describe and evaluate a planner that assembles components, described using this semantic model, into a dataflow graph. The planner also performs DLP reasoning to help determine connections between components.

This data has been imported from the ESWC2007 RDF