9
Initial Data Collection

Model building requires data. If you have data collected already, the data can be used to populate the MTR tables or the MTR schema can be mapped to the existing data via views. However, if you have no data, you might want to use the REAPI methods addItem and addItems to collect data. Data collection occurs in the Oracle9iAS Personalization Recommendation Engine (RE). For an RE to be up and running, there must be a package deployed in that RE. However, in order to build and deploy a package, you must have data in the MTR. To put it simply, you can't collect data unless you have enough data to build a package. You resolve this problem by populating the MTR with seed data and then using the seed data to build and deploy an initial package.

You create seed data on the system where Oracle9i is installed.

Creating and Removing Seed Data

OP includes a script to populate an MTR with seed data and also another script to remove the seed data from the MTR once enough data is collected in the MTR The following two scripts are in the directory $ORACLE_HOME/install/dbscripts:

• insertMTRSeedRecords.sql: Populates an unpopulated MTR with seed data.

• removeMTRSeedRecords.sql: Removes the seed data from the MTR once the initial model is built.

The OP administrator can run these scripts when appropriate.

Preparing to Build Models with Seed Data

Follow these steps to populate the tables required for model building; many of these steps are described in Getting Started with Oracle9iAS Personalization:

1. Install OP with an unpopulated MTR.

2. Configure synchronization settings in the MTR_CONFIGURATION table appropriately.

3. Create an MTR connection and allow synchronization for the data source types that you intend to collect.

4. Log in using the MTR account that you created during install, and run insertMTRSeedRecords.sql.

5. Create package using the default selections.

6. Create an RE Farm with at least one RE in it. Use the advanced settings to specify the customer data synchronization interval.

7. Build the package.

8. Deploy the package to the RE Farm that you created in step 6.

9. Logged in using the MTR account that you created during install, run removeMTRSeedRecords.sql.

10. Populate or map the following tables with data for your application:
    • MTR_ITEM: This table is the site catalog. When the collected data is saved in the MTR, data corresponding to non-existent items are thrown away as part of the validation check. So, if the MTR_ITEM table is empty, the collected data is thrown away.

    • MTR_CUSTOMER: The customer-IDs are expected to be managed by the site. Hence the MTR_CUSTOMER table should be populated or mapped with at least the ID attribute before the data is written to the MTR. The data values for other attributes can be collected via the data collection API.

    • MTR_ATTR_NAME_TO_ID_MAP and MTR_BIN_BOUNDARY: These tables should be populated with the mapping information that maps attribute names to numeric identifiers and the bin boundaries, respectively.

      You must specify bin boundaries for all data source types that you plan to collect.

11. After these tables are populated the OP administrator must run the following script, which populates the MTR_ATTR_ID_BIN_BOUNDARY table:

    populate_mtr_attr_id_bin_boundary.sql

Now collect data from your web site. After you collect data for an appropriate amount of time, you can build models based on the collected data.