Collaboratively administrate empowered markets via plug-and-play networks.
Business Rules Automation
Author and automate business rule execution for complex decision-intensive business processes. Integrate any business application or data source and maintain rules, processes and procedures from an intuitive graphical front-end. Monitor, alert and report on compliance in real-time.
In this sample
In this sample, a Store stock replenishment decision is taken on the basis of real-time data feeds on Store footfall, weather conditions and school holidays. The combination of these unstructured data sources is used to illustrate decision-making when conditions fall outside the expected norm. The sample application demonstrates the following technologies / capabilities:
XpertRule Decision Author (Graphical Rules Design Environment)
XpertRule Node.js Runtime Environment (installed on a BlueMix Instance)
IBM Alchemy Vision (to count number of people in an image)
IBM Insights for Weather (for current weather conditions)
How it works
Cognitive services – enable us to extract meaningful data from unstructured and abstract sources such as photos, images, speech. In this case, IBM Alchemy Vision interprets the number of people in the store from photographs
Automating Business Rules
So the question now is how to interpret and automate this combination of data using applied knowledge. Our sample uses the following flow of logic and decision making:
The main tree performs image recognition to determine the number of people in the image, retrieves the external temperature and then calls the AutoOrdering tree to determine which item to order.
Knowledge Auto-induction from decision table – The main AutoOrdering tree can be built graphically or alternatively (as shown here) can be automatically built using Tree induction from a set of example decisions. The tree utilizes two sub-trees StoreFootfall and CurrentTemp which convert numeric inputs into symbolic inputs.
The main AutoOrdering tree decides which item to automatically re-order. The knowledge utilizes two sub-trees StoreFootfall and CurrentTemp to provide semantic representation of the knowledge