Master Data File Generation

The data that companies rely on for analyzing their procurement activities, making purchasing decisions, assessing product engineering and supporting their customers is often "chaotic." Frequently, it is unstructured, presented in different formats, buried within a myriad of enterprise applications and databases, and captured within free-text notes and descriptions. The lack of a uniform, enterprise-wide system for integrating, managing and sharing data results in inefficiencies and miscommunications, and directly impacts the ability to make important business decisions.

For example, a hospital clinician may perform a search for information about medical products available from different vendors and be presented with the following records:

Using this information in its raw form makes it impossible to determine that all three products presented in the table are actually the same suture; not only is Johnson & Johnson spelled in two different ways, but the manufacturer, Ethicon, is a division of Johnson & Johnson. How many searches will the clinician have to perform to locate an 18 inch black suture? The color "black" is mentioned in the last two records as "BLK" and "BLACK" and the length of 18" is described in 3 different ways. Additionally, none of the records has explicit mention of suture absorbability, yet nylon material implies that it is a non-absorbable suture. The lack of structure in this data makes product searches, grouping, and procurement highly inefficient.

To address this widespread data quality problem, XSB, Inc. has developed an automated Master Data File (MDF) generation process for producing clean, structured, consistent data that can be instantly accessed and queried by any business unit within the organization. The XSB Master Data File finds all identical items based on standardized item manufacturer names and part numbers. Product descriptions are classified to a common taxonomy and product properties are extracted and standardized according to a uniform format. This allows for the creation of a single searchable MDF record for a group of identical products; its properties are an aggregation of all known properties about that item. Each MDF group is further partitioned into MDF subgroups based on standardized item packaging for efficient price comparison.

For the example above, the MDF would standardize all of the manufacturer names to Ethicon thus identifying all three records as identical parts (since the part number, C584D, is also the same). The MDF would then standardize packaging type information enabling the hospital clinician to compare prices from the various suppliers to buy the item at the best price available. For more detailed analysis, the MDF record would be classified to the Sutures class in the hospital’s taxonomy and attributes from all three records would be extracted, standardized and aggregated as illustrated below:

The MDF record retains relations to the original records so that queries for "Nylon suture" would yield all three records shown in the example.

MDF standardization ensures that clean, standardized data is available across the entire enterprise and facilitates efficient searches, attribute-based product comparisons, spend analysis and a variety of other uses.