Show TOC Entering content frame

This graphic is explained in the accompanying text CRM Intelligence Connector Locate the document in its SAP Library structure

Purpose

The SAP CRM Intelligence Connector enables you to perform real-time analytical functions on applications based on data mining models and provides real time access to key performance indicators. With the Intelligence Connector, you can provide optimal recommendations and personalized services in real time to your customers interacting with the CRM application.

Prerequisites

Before the CRM application can execute a real-time task, you must perform the following actions.

·          Installed the J2EE engine and SDM. For more information, see SAP Service Marketplace at service.sap.com for CRM Web Application Installation Guide for Release 4.0

·          Installed the CRM Intelligence Connector on the J2EE engine connected to the CRM4.0 application server, For more information, see SAP Service Marketplace at service.sap.com for Intelligence Connector Installation Guide 4.0, SP01 and subsequent Upgrade Guides

·          To call the Intelligence Connector administration user interface from the SAP Easy Access Menu, you must make the Customizing settings choosing SAP Customizing ® CRM Analytics ®   URL to Intelligence Connector Workbench. This is the link to the J2EE engine.

·          Specify prediction or KPI tasks and related objects in the CRM Intelligence Connector Workbench

·          When you execute the real-time tasks, you must ensure that all the relevant connections specified for the servers involved in the processing are available

Process Flow

This graphic is explained in the accompanying text

Application API: The Intelligence Connector provides an application API for the execution of real-time tasks. The application provides the task with specific input data executes the task and returns specific output.

Real-time tasks can either be prediction tasks or KPI tasks.

Prediction Task: When you execute a prediction task, the prediction which is based on a data mining model, for example, Campaign Automation uses the application API to set the input values (gender and age) for the current customer. The application executes the task and the predicted value, for example, a flag indicating if the customer will respond to a campaign or not and the predicted probability of response are obtained. 

KPI Task: When you execute a KPI task, the KPI is delivered to the application. For example, the application Campaign Automation may use the API to find out the average response time of a customer in the previously executed marketing campaigns. The application uses the API to set the input values for the KPI and returns a KPI value such as average response time.

The application API is available as a Java API and a set of ABAP functions.

Configuration and Administration Workbench:  Before a CRM application like Campaign Automation can execute real-time tasks, you must configure and define the tasks. You can define and maintain all the objects required for the execution of prediction and KPI tasks. The objects that you need to maintain are server types, KPI sets, data mining models and applications. For further details, see Maintaining Objects in Intelligence Connector.

The Intelligence Connector consists of a prediction provider and a KPI provider

·          Prediction Provider: The prediction provider offers real-time prediction based on data mining models by the executing prediction tasks. Currently, the prediction provider supports real-time prediction for classification models, clustering models, and regression models. A model store in a local database is used to store the data mining models and all related metadata needed to execute real-time tasks. The prediction provider can connect to various data mining servers including a local prediction server. These data mining servers can be used for the deployment of data mining models and for the processing of the prediction. To make the mining model operative for real-time prediction, you must deploy the model and process it for prediction. For more details, see Model Deployment and Prediction Processing.

·         KPI Provider:  The KPI Provider delivers the KPIs to the application by executing the KPI task. The objects maintained by the KPI provider for the KPIs are the KPI sets. Each KPI set has key fields (for example, Customer Id) to identify the KPI and one or more KPIs (for example Average Response Time to a Campaign) with specific types. The type of a KPI indicates where the KPI will be read from. The KPI can be read from a local KPI cache in the CRM application server or an ODS or a query in SAP BW.

Model Store: The imported (deployed) data mining model from SAP BW is stored here.

Metadata Store: The local database contains the metadata store and stores all the metadata needed for the execution of prediction and KPI tasks. For example, metadata could include the connections to data mining servers, model metadata such as model fields or the real-time task specifications.

Local Mining Server: This is where the algorithm is stored.

 

Leaving content frame