SCALE

Scales the score according to the number of words in all the phrase ranges attached to the rule

Applies to

• ALL
• ANY
• EXPRESSION
• LINK *
• MAX *
• MIN *
• PARAGRAPH *
• PHRASE *
• SENTENCE *
• TEXT
• WITH *

Note rules marked with * apply the scale to the calculated score - those without just use the weight as the rules score (provided that children pass the appropriate test where applicable)

Does Not apply to

• CATEGORY
• COMBINE
• SKIP

Values

• “1” - Do scale
• “0” - Don’t scale (default)

Algorithm

Calculate the story density - This is the ( ( number of words found by the rule ) * 10000) / ( number of words in document )

Calculate the percentage change based on the following table:

i.e. if story density was 1 then the percentage would be 80% - if story density was 100 percentage would be 98%

Then the existing weight for the rule is adjusted by this percentage (if no weight attribute is present for the rule this will be 100)

Example

If we had a rulebase like :

  <any scale="1">
    <text data="simple"/>
    <text data="example"/>
  </any>
 

and applied it to a simple document

 This is a simple example document to demonstrate the scale attribute.
 

We would have:

Number of words found by any rule = 2
Number of words in document = 11
Story density = 1818
Percentage = 100%
Resulting weight = 100
Since this is an any rule we are applied to this ignores the child scores and resulting score of rule would be 100

with a weight applied to the any rule we would have:

  <any scale="1" weight="50" >
      <text data="simple"/>
      <text data="example"/>
  </any>
 

If our example document had 2,500 words but still only 1 occurrence of simple and 1 occurrence of example then we would have:
Number of words found by rule = 2
Number of words in document = 2,500
Story density = 8
Percentage= 81
Resulting weight = 42
Again this would be the resulting score for an any rule (for other rules this may be weight * (child scores))

See also

• FOREACH