As pointed out in many of the previous articles on K-Praxis (XBRL and ebXML: Framing New Rules for Information Sharing and Business Intelligence: An Information Management Perspective) it is important to include unstructured and semi-structured data to obtain a complete 360-degree-view of organizational information. We have also been noting that there is a perceivable change in thinking of enterprise software vendors who are attempting to offer a more broader business analytics and decision support outlook based all the information that is available for processing.
Examples of Documentation Involved in Supply and Procurement Systems
Requisitions; Purchase orders; Procurement related documentation; Invoices; Travel and expenses documentation; Materials management documentation; Contract Documentation; Supplier related documentation; Credit Application and Credit Decision Related Documentation; Documents related to External Credit rating and Credit Storing processes; Customer Documentation; Billing including re-billed documents, project billings, billing complaints, shipment documents; Collections management documentation; Deductions management documentation.
Unstructured Text Information Locked in Human Interactions round Supply and Procurement Documentation
Examples of documentations provided above consists of basic generic documentation and many of these documents contain structured data (e.g., invoices, purchase orders, etc.) that can stored and accessed through databases and is available for reporting and analytics. But what goes untapped is information that is entered by humans in form of interactions, decision points, comments, text-based analysis (e.g., comments on purchase orders, decisions why a certain item was NOT purchased or comments about behavior of a supplier, etc). This unstructured or semi-structured information is not available in the databases and if even if it is entered text-field remains unused in future decision making processes.
And now software solution vendors have started providing solutions to help companies solve this problem. These solutions can be integrated easily into SCM or ERP systems of the organization and provide intelligent reporting and analysis of what is happening in the supply or procurement processes.
Technologies and Methodologies Used to Process Unstructured Information in Supply and Procurement Systems
At a meta level these vendors use natural language processing systems to tag the descriptive text data that lies around or inside the supply or procurement documentation. This data is aggregated from various formats, cleansed of unwanted text (stopwords), tagged, itemized, normalized, disambiguated and classified using various NLP methods (statistical and grammar based) in order to provide analytics for decision making purposes.
There are several companies providing these type of solutions. These companies include traditional ERP and SCM vendors like SAP,PeopleSoft and Ariba; Business Intelligence vendors like SAS and BusinessObjects; and more niche vendors whose main USP is to sell supply and procurement intelligence solutions. Among these niche vendors, most prominent are Softface, Zycus and VeriticalNet (Spend Analysis). These vendors are interesting to watch as they are attempting to infuse NLP and machines learning methodologies into core enterprise applications.