The OWLViper Project

    Overview

    The OWLViper tool merges OWL semantics, scientific analysis and data-mining with a user interface that makes it easy to build and visualize a complex set of processing routes that lead to a goal object. Basically, a Web Ontology Language (OWL) reader reads in an ontology (set of semantic linkages between terms) and then displays it in the GUI as menus of terms that users can chose from as they build their execution tree. The user selects classes of objects, the properties of interest and then puts constraints on property values. Also, there are transforms or operations that take in objects and put out other objects, and these also have properties and values that can be constrained. Working from the goal backwards to the source data in sequence, the user is guided by the application through the possibilities of available object transformations and data searches that can attain the goal.

    The fundamental principle which allows the Viper tool to work is that Viper attaches semantic meaning to both the types of data, and the operations/functions which transform those data. By doing so, it is possible to widen the users query of archive resources and to create a workflow of data transformations which are scientifically correct (as validated by the semantics in the overriding ontology). In the following example figure, a Tully-Fisher transform is used to convert source data into the target form the user desires.

    Viper allows for networks of transforms from many different source data repositories and intermediate data states. An example transformation workflow appears in the following figure.

    The Viper tool interface is still in development, and the sophisticated workflow shown in the preceding figure is not presently fully implemented. Still, some basic search/transformations are presently possible. A screenshot of the tool in its present form is shown below.

    Architecture/Components

    The Viper tool is comprised of a number of interacting components (figure 4).

    The Graphical User Interface (GUI) allows the user to initiate searches, fuse data and determine which semantic transforms are meaningful for their research.

    The Workflow Manager mediates the interaction between the other components and the GUI. It schedules when data are to be downloaded, and it manages the local database of retrieved and transformed data.

    The Data Agent provides the machinery to handle the needed queries and data retrievals from science archives. We have developed a framework using the Virtual Observatory Object Relational Mapping Layer (VOORML) which the Data Agent uses as its interface to various VO components.

    A layer diagram showing the interactions with VOORML in greater detail may be seen in the following figure.

    The Semantic Agent makes inferences from the ontology to enhance or facilitate the users requests. It comprises a JENA object layer coupled with + Pellet reasoner/inference engine to check for consistency with the inheritance tree and to determine what functions can be applied for input or output of stated objects.

    The Transform Agent allows the data be converted from one semantic form to another. We have developed a scientific data model used by the Transform Agent: the Collection/Quantity model which holds collections of scientific values bundled with important metadata, such as the definitions of the accuracy of the scientific value, and the associated units.

    Ontology. Presently we have developed and test Viper using an ontology of astronomy, however, the ontologies for other fields of science can be developed, and therefore OWLViper could be applied to other fields.

    Presentations

    A summary for our funding agency, NASA's AISRP program.

    The AAS 2007 Poster summarizing the Viper project.

    Release Schedule/Software

    Software is not presently available for Viper. We are planning to make an initial release of Viper in Spring 2008 and a link will appear on this website when that occurs. Please keep checking back here for updates to the status.


    Webmasters: Ed Shaya / Brian Thomas