LRC Topics

LRC Topics are Verity topics created using LRC. After an LRC topic is created, it is inserted in the current topic set in Business Console and displayed in the topic pane automatically. The figure below shows an example of an LRC Topic on “acquisition”. All LRC topics use the <LogSum/n> operator in the top-level named node, where the integer n is also automatically learned by LRC and can be different for different topics. The <LogSum/n> operator can have up to 1024 child nodes (evidence nodes). Each child node consists of a weight-evidence pair.

Weight is specified as a signed integer “w” in the <MULT/w> operator. An evidence node can take one of the following two forms:

<Word> SomeTerm

or

<Field> SomeField <Contains> SomeTerm



When an LRC topic is evaluated against a document, all the positive and negative evidences that appear in the document are accumulated according to the formula for the <LogSum> and <LogSum/n> operators. The score is always between 0.0 and 1.0. For more information about the VQL operators, see the Verity Query Language and Topic Guide.

 

 
Note   It is possible to have a non-zero score for a document that does not contain any evidence, but normally this non-zero score is close to zero.