Today, when a retailer needs to assemble a complete view of informa- tion about a product for an online product catalogue, that retailer will typically go to multiple web-based sources, search for information, transform the information obtained into a format fit for purpose, and assemble the information elements for publication to the catalogue.
Today that content integration process is difficult. It is difficult because the steps of the process and the technology required for search and retrieval of information vary sufficiently from source to source, and even from item to item such that it is difficult to fully automate. The content integration problem blocks operational efficiency, in- troduces product information errors, and slows down the speed to market - resulting in increased costs and lost sales.
One of the approaches traditionally proposed for solving this con- tent integration problem is “data federation”. Data federation works by standardising on a common federated data model and map- ping all data sources to that standardised model. Content might be mapped either real-time in response to requests, or it might be stored using the common data model in an intermediate data store ready for consumption.
The challenge though is that agreeing on a standard model is difficult and despite big efforts and successes in standardisation across the value chain there always seem to be exceptions and in most cases there remains data which does not fit the model. We’ve been asked to find a more general way to share information such that partners can c