Ensure the Absolute Quality of an Attribute

If the absolute quality is defined, it is used on new weight before comparing it with the existing data. The following example shows how to ensure the absolute quality of an attribute:

  1. John is an existing name in the system.
  2. A new value is received: J. The source from where J comes from is more reliable than the previous source.

In the current implementation, J overwrites John. You can have a rule which can reject this record but as this record contains other useful information, you do not want to reject it. You would rather just change J to John. This is where the concept of absolute quality of data comes in. In this example, the quality of J is less than the quality of John.

A capability to assign semantics and support it with a way to add rules for quality computation is needed. This can be done in one of the following ways:

  • When defining the precedence data for each attribute, attach a rule for absolute quality computation. This rule is fired when precedence is calculated.
  • Have an action in the rulebase, assignQuality, which assigns a numeric value attribute quality. This value is retained and subsequently used for precedence calculations.

  • Similar to the first option, but instead of defining a rule per attribute, define a rulebase which computes absolutequality for all of the attributes.

When absolute quality is computed, it is multiplied by the source trust factor to arrive at the applicable quality of the data before comparing it with the existing data.