This section provides an overview of the proposed architecture, which includes a set of core services and tools. Section 5.2 - Overview of Core Semantic Infrastructure Capabilities and Services Profile summarizes the profile of the solution with mapping to appropriate requirements and use cases. Section 5.3 - Tools for Semantic Infrastructure 2.0 provides an end user's view of the tools. Section 5.4 - User Workflows in Semantic Infrastructure 2.0 describes workflows, and section 5.5 - Tie-in with Terminology and Platform describes integration with the platform and terminology.
The image below gives an overall view of the components required for Semantic Infrastructure 2.0.
Note
A description of the relationship among components in the diagram will be provided.
Legend for Diagram of Architecture of Semantic Infrastructure 2.0
Component Name |
Description |
|---|---|
Box name:Metadata Management |
Provides for the expression, capture, storage, and retrieval of metadata, including models, datatype usage, terminology requirements and contextual aspects of model components (Note: The Metadata and Model Repositories logically reside in this area. This varies from but is in line with the solicitation as all repositories will be developed.) |
Box name:Metadata Stores |
Is concerned with providing context and meaning to data and will include data elements, terminology references, and informational knowledge. |
Box name:Model Management |
Models constitute the expression of metadata elements for a particular domain or problem set. Model Management allows elements to be collected and related to provide for useful solutions. Models are described using OMGs XMI language, providing for the foundations of common tooling and expression. As important as XMI, are the human-readable models expressed in familiar UML. These models contain essential elements for solutions to function. |
Box name: HL7 v3 RIM meta Infomation |
The precision afforded by HL7 V3 is afforded by models that are used to characterize data as well as metadata for a model's elements. Managing these models as Metadata allows use and reuse in the day-to-day operations of NCI, allowing precision and accuracy in exchanging data within the BIG Health scope |
Box name: Knowledge Management Service Bus |
The Service Bus allows for a versatile, flexible approach to providing additional functionality that leverages the existing components of the semantic infrastructure. It is intended to leverage and extend other service bus solutions in service at NCI and to allow the integration of federated artifacts into NCI administered semantic infrastructure space. The Knowledge Management Service Bus exposes direct artifact retrieval as well as discovery services to identify knowledge components that may be of interest to applications. (Note: The Knowledge Service, Metadata Repository(MDR), and other services reside here and we consider the Model Service logically part of the MDR Service.) |
Box name:MDA Services |
Provides for an extensible platform based upon semantic web technologies to support registration, search and retrieval of models within the knowledge repository. |
Box name:Knowledge Services |
Provides for an extensible platform based upon semantic web technologies to support registration of, search for and retrieval of knowledge components and the analysis of such components based upon their metadata characteristics. |
Box name:Transformational Services |
The Transformational Services provide for an extensible platform, based on Semantic Web Technologies to support enterprise initiatives to leverage deployed data and metadata by providing tools to extend, scale, or query existing data stores. |
Box name:Knowledge Management |
Provides for the expression, capture, storage, and retrieval of knowledge concerning usage of metadata and data in the deployed architecture. Knowledge includes concepts and usage, both of which may be expressed using formal ontologies. |
Box name:Semantic Knowledge Store |
The Semantic Knowledge Store leverages RDF Triple Stores to provide for concepts and their relationships. |
Box name:Knowledge Analysis |
Knowledge Analysis provides for the rules and the reasoning engines that leverage knowledge captured in the Knowledge Management stores to extend scientific and clinical research, as well as provides the basis for an extensible rule-driven alert network. |
Box Name:Terminology Management |
Allows for the expression, storage, and retrieval of terminologies and vocabulary elements to support models, usage, metadata, and knowledge management. |
Component Name |
Description |
Box Name : Artifact Reference |
The artifact reference is a store or registry that contains references to the various artifacts. Each artifact should have a URL that can be used to physically access the file. Each artifact reference is accompanied by a checksum or some other method to be able to verify the accessed object. |
Box Name : Artifact Access |
This service provides a programmatic method for accessing artifacts within the SI. |
Box Name : Transformation |
This service provides a service that has a number of services that take as input some artifact and outputs in alternative representations. This might include a class model in UML being transformed to an OWL ontology. |
Box Name : Semantic Knowledge Store |
This store provides a physical representation of semantics that have either been derived through artifact analysis, or through manual annotation. This store could be represented by an RDF triple store. |
Box Name : Artifact Registry and Retrieve |
The registry and retrieve service provides a programmatic interface for interacting with the artifact reference registry. |
Box Name : Automated Semantics Discovery |
This service takes an artifact and extracts as much semantics as possible. The details and amounts of semantics will depend on artifact type, representation, and completeness. |
Box Name : Annotation |
The annotation services provides functionality which allows additional semantics to be added to an artifact reference. |
Box Name : Orchestration |
The orchestration service manages the internal flow of operations which can be performed. This includes automating the transformation and semantic discovery and the utilization of various rule systems or classification systems. |
Box Name : Governance Integration |
To support governance, this service provides state mechanisms about known artifacts that can be accessed and reviewed through governance activates. |
Box Name : Access Service Directory |
This directory represents the set of services that are available within an SI implementation which are designed to manage artifacts and their semantic representations. This will allow for the coordination of stores and services across the grid. |
Box Name : Rule Systems |
The rule systems provide integrations of one or more rule systems that provide support to the SI to express business rules and behaviors. |
Box Name : Classification Reasoners |
Classification reasoners provide integration to one or more classification tools. These tools are systems that process semantic information and dependant to determine relationships and associations of classes and individuals which may be expressed in an artifact, its annotated information, or instance representations of associated data. |
Box Name : |
The expert systems interface provides integration to one or more expert systems. These systems utilize a set of known facts and definitions to determine additional semantics and functional definitions within the artifact semantic information and instance representations of associated data. |
Box Name : Reasoning Framework Service Directory |
This directory represents the set of services that are available within an SI implementation provide reasoning functionality to analyze artifacts and instance representations of associated data. |
Box Name : caGrid 2.0 |
Is the connectivity and secure transmission hub for communications with institutions utilized by NCI and it’s associated cancer centers, research centers, and affiliated organizations. |
Box Name : Grid Application Toolbox |
This is a collection of tools and libraries which are designed to make integration to the caGrid easier and more efficient. |
Box Name : caGrid Enabled Applications |
Any application that utilizes the grid for communications. These apps may utilize the Grid Application Toolbox or provide their own interface to the caGrid . Examples of these applications include the caBig Clinical Information Suite and caTissue. This will also include infrastructure tools such as Form definition tools, query tools and code generation systems. |
Box Name : Semantic Annotation Application |
This application is a caGrid enabled application which provides users with the ability to annotate artifacts in an SI framework implementation. This application is likely a Web Based application that may be part of the caGrid Portal. |
Box Name : caGrid Portal |
The caGrid Portal is an application that provides support of the integration of grid components. From the portal identification of services and data is performed to expose that information to the other users of the grid. |
Box Name : Clinical Data |
This represents clinical information that may be exposed to the grid. Using the portal, an authorized user may expose data or services onto the grid, this might include outcome markers, treatment plans or other relevant information |
Box Name : Clinical Research Data |
This represents clinical research data that might be exposed to the grid. Using the portal, an authorized user may expose data or services onto the grid, this might include trial cohort qualifications, raw data, or publishable results. |
Box Name : Life Sciences Data |
This represents life sciences data that might be exposed to the grid. Using the portal, an authorized user may expose data or services on the grid, this might included gene array studies, algorithms, methodologies and data sets. |
An example problem:
A researcher wishes to collaborate with another researcher to more precisely define a treatment plan for some individuals. He believes the best way to do that is to expose some data that he is collecting to the other researcher. This information is changing and expanding, and so merely sending a dataset is insufficient.
How the architecture supports the solution:
If the user has followed an expressive methodology (such as ECCF) to design his dataset :
Using the caGrid portal, the user logs in and indicates that he wishes to share a dataset with another researcher. This dataset is accessed via a database, so he must connect the database to the caGrid. The user will have at his disposal various artifacts that describe his data and it’s representation. If he has not done so before, he will register the appropriate artifacts (models, specifications, etc) using the caGrid Portal.
For each artifact, the Semantic Infrastructure services are used to perform an orchestrated flow to learn about the artifact. First the system will register the artifact in the si, the si will access the artifact and perform any transformations as necessary to most effectively process it’s contents. The set of rules, classification definitions and expert systems are utilized to extract semantics such as the set of problem domains the information represents, the mood of the information, and specific elements such as standards of coding schemes and value sets which are used by the system. All this information is stored in the knowledge store using common semantics used to define datasets.
Or, if the user doesn’t have those artifact available :
If the user does not have artifacts, he isn’t out of luck. By providing access to the data resource the system will attempt to determine aspects of the data looking at the representation in the data itself. Since many systems are not self-describing, the user may need to provide more information during the annotation process.
Providing a dataset link :
The user is now ready to provide a link to the grid using caGrid portal. The user provides the parameters that are required to connect and access the dataset. These parameters are different depending on the type and presentation of the dataset, but generally the user will provide a URL or database connectivity information. Assuming there are no artifacts the system will generate artifact representations of the data set so that the user can annotate the data.
Annotating the artifacts :
In some cases, the user may have to annotate aspects of the artifacts that represent his data. This is done in situations where there are not supporting design models. User will provide general metadata that describe the problem domain and specific data representations used. This may include code system and value set use, and other elements which effectively document the dataset. Generally the more the user annotates, the better the consumers of his data will be able to access and query his data.
Done.
The user has placed his data set on the grid in a way that can be accessed by authorized users.
