Overview of correlation
Correlation is the process of associating multiple pieces of data with each other based on strong identifiers. For the process of ingesting data into the Information Store, i2 Analyze can use correlation identifiers that you provide to determine how to process and represent data in i2 Analyze records.
Correlation in i2 Analyze
During the ingestion process for the Information Store, correlation can be used to determine when data that is ingested is associated with existing records, and must be represented by a single i2 Analyze record. Conversely, if data no longer represents the same object, the data can be represented by multiple i2 Analyze records. In i2 Analyze, these operations are known as merge and unmerge.
During the correlation process, an identifier is used to determine how each row of data is associated. You present the identifier to the Information Store with the other staging data during ingestion.
You can limit the use of correlation to data from a specific source, of certain item types, or per row of data ingested into the Information Store. You do not have to provide a correlation identifier for all the data that you ingest into the Information Store.
You might want to use correlation when you are ingesting data that originates from disparate sources, which have common properties, or have the potential to represent the same real-world objects. For example, if you have two data sources that contain information about people.
Another scenario where you might use correlation, is when the data that you are ingesting is in the form of event driven models (crime or complaint reports) where the same actors (people, locations, phones, and vehicles) might be referred to frequently in the same source.
Correlation can be used in these scenarios to combine multiple source records into single i2 Analyze records for link analysis.
In i2 Analyze, correlation identifiers and implicit discriminators are used to determine how the Information Store processes data during ingestion.
When you ingest data into the Information Store, you can provide a correlation identifier type and key value that are used to construct the correlation identifier for each row of data in the staging table. The type and key values that you provide are used to process data that is determined to represent the same real world object. Implicit discriminators are formed from parts of the i2 Analyze data model in the Information Store. Even if correlation identifiers match, if values for elements of the i2 Analyze data model are not compatible, that data cannot be represented by the same i2 Analyze record. For more information about correlation identifiers and implicit discriminators, see Correlation identifiers.
During the ingestion process, i2 Analyze compares the correlation identifiers of the data to be ingested and existing data in the Information Store. The value of the correlation identifiers determines the operations that occur. For more information about the correlation operations that can occur, see Correlation operations.
The example data sets demonstrate the correlation behavior in this release of i2 Analyze. For more information, see Ingesting example correlation data.