How Linked Data Impacts Enterprise Data Access, Integration, and Management -- Part 1

As part of an age-old endeavor to exploit IT en route to attaining and sustaining optimal enterprise agility levels, enterprises have acquired a plethora of line-of-business (LOB) applications (apps) from a cocktail of vendors. Sadly, the perpetual pursuit and acquisition of these apps inadvertently leads to an explosion of underlying data management systems across the enterprise, that vary across the following dimensions:

1. Vendor

2. Underling Data Model

3. Query Language Support & Compliance

4. Data Access Protocols & Compliance

5. Supported Operating Systems.

As a result of these issues, its extremely common to find enterprise data silos (or islands) spread across operational domains such as Accounts, Human Resources (HR), Customer Relationship Management (CRM), Supply Chain Management (SCM), Corporate Communications, etc.

Enterprise Linked Data, Demystified

This post provides a simple guide showing how Linked Data profoundly affects (in a positive way!) the dynamics of enterprise data access, integration, dissemination, and management.

Situation Analysis

Often lost in the conversation about IT is the fact that for the entirety of the modern IT era, there has been strong desire on the part of IT decision makers to crystalize some variant of the vision easily recognized as "Information At Your Fingertips." To date though, that vision has remained mercurial, at best.

The Value Proposition of Enterprise Linked Data

As a contemporary InterWeb-era mechanism for Data Connectivity, Access, and Integration, equipped with the full prowess of HTTP and REST, Linked Data addresses several challenges common to all enterprises:

1. Discovery, Manipulation, Dissemination, and Storage of relevant Data, Information, and Knowledge.

2. Loose coupling of Data, Information, and Knowledge.

3. Reconciling identical records across disparate databases -- due to typos, keys uniqueness scoped to line-of-business applications oriented data silos

4. Network as a truly functional computing device without boundaries.

5. Attaining all of the above without compromising security and privacy.

Creation, Deployment, and Exploitation of Enterprise Linked Data

Proposing to rip-and-replace existing infrastructure is a non-starter when it comes to incorporating new technological advances. This is non-negotiable at the enterprise level. Thus, exploitation of Linked Data has to be driven by dynamic generation from existing data sources, which will always be associated with a plethora of heterogeneous data silos.

In a world where data source heterogeneity is a fact of enterprise life, data virtualization provides the most compelling and practical entry point for exploitation of Linked Data.

What does Linked-Data-enhanced data virtualization deliver?

The following courtesy of hyperlinks that leverage URI abstraction:

1. Every data item, record, entity is represented by a Data Object.

2. Every Data Object is unambiguously identified using an Identifier the serves the role of Name or Handle -- resolvability (ability to de-reference) of these Identifiers is optional, but highly recommended.

3. Every Data Object is a Resource that consists of an Entity-Attribute-Value (EAV) based graph pictorial, that facilitates actual Object Representation at the time of expression or transmission-across-the-wire via serialization.

4. Every Data Object is accessible via an Address.

5. Data Object Names and Addresses are Distinct even when they are Hyperlink-based.

6. Every Data Object is serializable across-the-wire using a variety of negotiable data serialization formats.

As a result of these fundamentals, a number of data access and integration challenges become solvable:

* Equivalence -- covering co-reference and value-based-equality (i.e., two unambiguously-named objects could share a co-referent even as their EAV graph pictorial-based representations are constituted of utterly distinct values).

* Locale Context Variations -- different users of data perceive, communicate, and measure using a variety of default symbols covering languages, units, and metrics.

* Separation of Data Access Protocol from Data Representation Syntax and Data Serialization Formats.

* Orientation of data access and modeling on the Conceptual Level, rather than the Application Logic level.

* Separation of Conceptual Schema from Data Representation Syntax.

* Late and Loose bindings to Conceptual Schema(s).

* Cumulative Contextual Fluidity -- different people (beyond locale variations covered above) require a variety of views (context lenses, so to speak) over the same data, subject to a myriad of data interaction factors (e.g., position, time, location, etc.).

* Context-aware Data Access policies -- context-sensitive rules governing data access.

Example

This example covers the generation of Linked Data views over an RDBMS oriented data source. Access to the data source is achievable via ODBC, JDBC, or shared locality in a hybrid mutli-model DBMS server. In the case of Virtuoso [1], you can choose which of the aforementioned options best serves your needs.

Relational Schema (Data Dictionary)
Northwind (demo DBMS for SQL Server and OpenLink Virtuoso) which is comprised of the following tables:
* Orders
* Products
* Suppliers
* Shippers
* Countries
* Provinces
* Employees
* Categories

IRI for the Named Graph containing the Ontology (TBox generated from Relational Schema) Data
<http://demo.openlinksw.com/schemas/Northwind/> .

IRI for the Named Graph containing the Instance (ABox) Data
<http://demo.openlinksw.com/Northwind#> .

IRI for the Dataset Metadata (VoiD Graph)
<http://demo.openlinksw.com/Northwind/stat#>

Sample URIs from the Instance Data URI
1. http://demo.openlinksw.com/Northwind/categories/CategoryID/1#this -- an object identifier for "product category" object
2. http://demo.openlinksw.com/Northwind/countries/Name/Aruba#this -- an object identifier for a "country" object
3. http://demo.openlinksw.com/Northwind/customers/CustomerID/ALFKI#this -- an object identifier for a "customer" object
4. http://demo.openlinksw.com/Northwind/employees/EmployeeID/3#this -- an object identifier for an "employee" object
5. http://demo.openlinksw.com/Northwind/orders/OrderID/10258#this -- an object identifier for an "order" object
7. http://demo.openlinksw.com/Northwind/products/ProductID/1#this -- an object identifier for a "product" object
8. http://demo.openlinksw.com/Northwind/shippers/ShipperID/1#this -- an object identifier for "shipper" object

SPARQL Endpoints
1. http://demo.openlinksw.com/sparql
2. http://demo.openlinksw.com/isparql .

SPARQL-FED endpoints
1. http://uriburner.com/sparql
2. http://uriburner.com/isparql .

Related

1. http://virtuoso.openlinksw.com -- Virtuoso Home Page.

#LinkedData #BigData #NoSQL #DBMS #RDBMS #ODBC #JDBC
PhotoPhotoPhotoPhotoPhoto
22 Photos - View album
Shared publiclyView activity