Policy Guide › Document Classifications › Parameter 6 Functions

Parameter 6 Functions

You can add these functions to parameter 6 to modify the document score handling:

Important! ReduceBySize(r) and Normalize(m) are alternative methods for adjusting document scores downwards. Do not include both functions together when specifying parameter 6.

MinScore(n) / MaxScore(n)

These define the minimum and maximum document scores needed for CA DLP to confirm a document classification. They can be used individually:

• Enter MinScore(n) where n is the value. For example, type MinScore(10) to set a minimum score of 10.
• Enter MaxScore(n) where n is the value. For example, type MaxScore(10) to set a maximum score of 10.

Or, used together, MinScore(n) and MaxScore(n) can specify a set of ranges for document classification scores, or ‘severity bands’. For example:

Classifier 1

Parameter 3: %CCN%

Parameter 6: MaxScore(10)

Severity: Low

Classifier 2

Parameter 3: %CCN%

Parameter 6: MinScore(11),MaxScore(100)

Severity: Medium

Classifier 3

Parameter 3: %CCN%

Parameter 6: MinScore(101)

Severity: High

Where:

Classifier 1 specifies that detecting less than 10 credit card numbers qualifies as low severity.

Classifier 2 specifies that detecting 11-100 credit card numbers qualifies as medium severity.

Classifier 3 specifies that detecting more than 100 credit card numbers qualifies as high severity.

ReduceBySize(r)

This function adjusts the document score downwards in order to prevent false confirmations, especially for long documents such as email attachments or uploaded files.

Enter ReduceBySize(r) where r determines the value subtracted from the document score; r is typically a very small value, say, 0.3. The formula for this function is:

NDS = ODS - (Characters * (r/100))

Where:

Characters is the number of characters in the document.

NDS is New Document Score

ODS is Old Document Score

For example, if the original document score is 45, and the document size is 10,000 characters (a typical score for a four or five page document), then setting r to 0.3 lowers the document score to 15:

15 = 45 - ( 10,000 * 0.003 )

Note: This function can result in fractional scores. Such scores are rounded to the nearest integer before testing against MinScore(n) or MaxScore(n).

Normalize(m)

This also adjusts the document score downwards in order to prevent false confirmations. It is an alternative method to ReduceBySize(n).

Enter Normalize(m) where m determines the multiplier used to lower the document score; typically, m matches the average size (in characters) of the documents you want to classify, for example, 10,000. The formula for this function is:

NDS = ODS * (m/Characters)

Where:

Characters is the number of characters in the document.

NDS is New Document Score

ODS is Old Document Score

For example, if the original document score is 45, and the document size is 25,000 characters, then setting m to 8,500 would lower the document score to around 15:

15.3 = 45 * (8,500/25,000)

Note: This function can result in fractional scores. Such scores are rounded to the nearest integer before testing against MinScore(n) or MaxScore(n).

NotSmallerThan(x)

This function defines a minimum document length. CA DLP does not attempt to classify documents with fewer than x characters (excluding white spaces and characters such as carriage returns). This is useful if you know the typical size of your target documents. For example, to ignore documents smaller than 500 characters, enter:

NotSmallerThan(500)

You can use this function in conjunction with its NotLargerThan(y) counterpart to define the permitted size range for target documents.

NotLargerThan(y)

This function defines a maximum document length. CA DLP does not attempt to classify documents with more than y characters (excluding white spaces and characters such as carriage returns). This is useful if you know the typical size of your target documents. For example, to ignore documents larger than 2000 characters, enter:

NotLargerThan(2000)

You can use this function in conjunction with its NotSmallerThan(x) counterpart to define the permitted size range for target documents.

Extract(1,2,3,4,8)

This function causes keystrings found using the classification parameters 1, 2, 3, 4 and 8 to be extracted and saved as an attribute of the capture or control event. You can extract the keystrings found using any or all of these classification parameters. For example, if you are only interested in the keystrings found using parameters 1, 2 and 8, enter:

Extract(1,2,8)