Semantic Source to Pay System - The Holistic Approach to Enterprise Software
The foundation of data model is one of the most critical factors contributing to the success of any enterprise software system. A well define and extendable data model will greatly improve the evolution and maintainability of the system.
The most prevalent technology for the enterprise software data design is the combination of the RDMS for the under hood data model and XML for the integration. Besides all the advantage of relational DB scheme and XML schema, two of the biggest problems come with such models are:
1). Tight coupling of data model with the application.
2).tremendous data redundancy for storage and integration.
Thinking about the typical Source to Payment software suite, the data model of the contract lifecycle management (CLM) application is so coupled with the application itself that it could never be used for the procurement application even though most of the procurement applications need the exactly same contract data for their business process management process. Many times a redundant and similar table structure is duplicated into the procurement application. To make the things worse, a sync-up process between two will produce more redundancy and waste of resources. Another example would be the transactional business data: in each purchase order, change order, order response and invoice, the buying/selling/billing parties' address, communication means and identifier are repeatedly transported and stored between procurement application and backend ERP system even though 99% time they are purely duplicate and redundant data. The root of the problem is exactly I pointed out above that the rigid data model and schema are forcing data itself unfortunately intertwined with the process (application).
Someone would say, while I get your point but that is the price we have to pay to tradeoff the flexibility with the efficiency. The answer is yes and no; yes, that was the price a system had to pay when there was no mature technology to replace the intrinsic rigid nature of relational data model and schema; no, fortunately nowadays we are a "secrete" weapon in our arsenal to tackle the above problems of undesired coupling between data and application and unnecessary data redundancy between applications.
That holy grain is the semantic data approach. The semantic model is a simplistic but very powerful way to present the data; it comes with three parts subject-verb-object. Virtually all data and relationship would be deduced to these three elements: Infosys has a master agreement A; the master agreement A is for Secure ID; the Secure ID is produced by RSA; RSA is a US company; RSA HC's address is in MA; RSA's plant address is in TX; the PO has a number of 123; PO number 123 contains 100 RSA secure ID.
As you can see that any complex relationship or transactional data could be bisected into semantic statements which frees up the data model completely from the application and an advance infer engine with the properly defined ontology would understand and harvest the information from the semantic raw data by taking advantage of powerful 64-bit O.S. and in-memory DB, the true power of semantic data anchored applications are finally ready for the prime time. Taking enterprise Source to payment domain, imagine a common data store back by the semantic data model (RDF and S2P ontology) there will be no extra data transformation and integration between spend analytic and procurement application since all the data is stored in the semantic form and spend analytic application just need to refer to the S2P ontology and pre-defined rules to extract and analyze; the only information for the flow of contract from strategic sourcing to procurement is the master agreement/contract ID, the procurement system will be intelligently enough to extract the related contract information using the semantic data on the fly when it is necessary during its workflow process; finally there will be no more redundant data in PO, CO, POR, INV and GR passing around, what any application need now is the true delta; with the help of the infer engine and ontology the application would extract precise information as required for its process Just-In-Time.
It is not only the tremendous efficiency and resources saving would be exponential comparing to the current mainstream systems but also the foundation of the computing system which is much more close to the way human brain store and handle information would be established and prepared for the next revolution. Combining the big data with semantic data model smarter enterprise is finally not maybe but INDEED J