Workflow-Driven Ontologies: An Earth Sciences Case Study

Abstract

A goal of the Geosciences Network (GEON) is to MotivationThe NSF-funded Geosciences Network (GEON) project is a collaborative effort among researchers from a broad cross section of computer science and earth science disciplines [1]. GEON is comprised of computation clusters that include compute nodes and data nodes that store high quality geological information and software services and that enable data access, analysis, modeling, and visualization. GEON is an example of a "virtual scientific community" that uses cyber-infrastructure (CI) to support and enhance the scientific process.Different groups from the GEON community are developing ontologies that support search [2], information integration [3], and service discovery [4]. An ontology [5] is an explicit specification of the objects, concepts, and other entities that are assumed to exist in a specific domain and the relationships that hold among them.In addition to development of ontologies, many research teams are using workflow techniques to specify the computation of complex scientific activities [6,7,8]. In this paper, we describe a new approach to ontology design called Workflow-Driven Ontologies (WDO). WDOs are distinguished from domain-based ontologies that capture basic knowledge about a domain. Use cases typically drive the specification of domain-based ontologies [9]. In the WDO approach, abstract workflow specifications drive the elicitation and specification of classes and their relationships. For example, domain experts, i.e., earth scientists, begin the knowledge acquisition process by identifying a product and from the product identify methods that can generate the product. Further, domain experts can identify data that are required as input for the identified methods. Knowledge acquisition methodologies based on WDOs are flexible since earth scientists can refine WDOs by refining a WDO-derived workflows and vice-versa. We claim that abstract WDO-derived workflow specifications are indeed the use cases for WDOs.Prior to presenting the details of WDOs, this paper first motivates the utility of WDOs by presenting a case study in Section 2. The case study illustrates how a contour map can be generated from composition of services from the gravity domain. Section 3 explains how workflows are derived from WDOs using the WDO class hierarchy and core relationships. Section 4 presents related work including a discussion on how WDOs compare to other ontologies. Section 5 summarizes the main contributions and open issues related to the development of the WDO approach. Gravity Case StudyThis section describes the Gravity ontology [10] and the use of the ontology to specify workflows. To remain consistent with the terminology used by the OWL Web Ontology Language community, we use the term "class" to denote types of objects captured by the ontology.

Program synthesis from Workflow-Driven Ontologies

“…(1) A + B + C = π, i.e., the sum of the angles is 180º, or π radians; Suppose that the side lengths a, and b and the angle A are known, and it is necessary to determine the length of side c. As a result, the knowledge in this example problem is represented by relations (1), (2), and (3) and the values of a, b, and A; and the unknown value is c.…”

“…Our previous work has resulted in the development of the Workflow-Driven Ontologies (WDO) approach [2]. The intention of WDO is to facilitate the capture of process knowledge by scientists directly with terms from their own domains of expertise.…”

“…This concept and relation form simplifies the process of automating workflow generation by uniquely identifying the Information concepts related to a given Method on a workflow specification. The automatic generation of workflow specifications from WDO ontologies was initially explored in [2], and it is formalized here through the application of established program synthesis techniques.…”

“…As a consequence, scientists now need assistance with integrating these new technologies into their practices. Salayandia et al in [2] describe the Workflow-Driven Ontologies (WDO) approach that has been developed with the intention of enabling domain scientists to create and customize their own applications by composing resources that are accessible over CI. A WDO consists of an upper-level ontology that can be extended by domain scientists to capture process knowledge or knowledge about how scientific tasks are conducted.…”

Knowledge Annotations in Scientific Workflows: An Implementation in Kepler

“…WDOIT! [8] is a scientific workflow tool that publishes all workflow information as semantic annotations on the Semantic Web. With these annotations, a WDOIT!…”