Semantic Maps

semantic Map

A resource for the life science corresponds to an information source such as a data repository or database management system (e.g., a query form or a textual search engine), a link between resources (an index or hyperlink), or a service such as an application or tool. Resources are characterized by core information including a name, a description of its input and its output (parameters or format), its address, and various additional properties expressed as metadata.

Resource discovery is the process of identifying and locating existing resources that have a particular property. Machine-based resource discovery relies on crawling, clustering, and classifying resources discovered on the Web automatically. Resources are organized with respect to metadata that characterize their content (for data sources), their semantics (in terms of ontological classes and relationships), their characteristics (syntactical properties), their performance (with metrics and benchmarks), their quality (curation, reliability, trust), etc. Resource discovery systems allow the expression of queries to identify and locate resources that implement scientific tasks.

The Semantic Map project aims at developing technology to support resource discovery.

project

The Semantic Map projects include:

· BioNavigation

First prototype developed by Kaushal Parekh (ASU) with Zoé Lacroix, Maria Esther Vidal, and Louiqa Raschid.

· Semantic Map for Structural Bioinformatics

Second prototype developed by Hervé Ménager (ASU) with Zoé Lacroix and Pierre Tufféry.

· Semantic Map for BioMoby services

Evaluation in progress conducted by Maliha Aziz (ASU) with Zoé Lacroix, Hervé Ménager, Pierre Tufféry.

Return to Scientific Data Managament Lab

emantic Map for Structural bioinformatics

Semantic Map for Structural Bioinformatics

Challenges

(a) All existing resources for a relevant to a particular domain need to be represented in the semantic map. This is a challenge because of the magnitude and complete lack of regulation of bioinformatics development. There are thousands of bioinformatics resources including data sources and tools publicly available on the Web. To address this problem, the Semantic Map approach aims at developing an automatic registration process and a customized view mechanism to retrieve and display resources.

(b) There is not a commonly agreed format to publish a resource. Recent efforts such as Web Services and BioMoby aim at homogenizing resource publication. Although they improve resource interoperability, they yet lack the semantics that is needed to locate a resource with respect to what it achieves or implements as needed for resource discovery. To address this limitation the Semantic map approach uses a domain ontology to classify bioinformatics resources.

(c) Resource discovery is needed when scientific protocols (workflows) are designed. From a design protocol that captures the scientific aim (in terms of concepts and relationships of an ontology) a scientist needs to identify resources that implement each of the conceptual task. Scientific protocols are directed graphs (see ProtocolDB for a discussion on scientific protocols and their structure). The identification of a bioinformatics resource suitable to implement a scientific task relies on the validation of three constraints: (1) does it capture the expected semantics? (is it expected to achieve the intended task?), (2) can it be connected to the previous and following resource? (interoperability or syntactic validation)?, and (3) will it produce an efficient workflow? (performance). To address this issue a query language will be developed to identify resources semantically suitable and BioOnMap approach will be used to validate the interoperability constraint. Solutions developed in the context of BioFast and BioNavigation will be used to produce efficient implementations.