cancel
Showing results for 
Search instead for 
Did you mean: 

How to measure the classification rules performance?

Joao_Salgado
Not applicable

The Veritas Information Classifier was designed to be able to work with many rules simultaneously but poorly performant individual rules may affect the overall performance of classification. In order to easily identify which rules perform better or worse, the Veritas Information Classifier can output the list of rules and their performance.

 

DTrace

DTrace will output the performance of all of the rules and the percentage of time that was spent on each of them.

  1. Enable DTrace for fsdmhost and start logging
    DT> Set fsdmhost v
    DT> Log
  2. Set specific filters in DTrace that will allow to easily review the content
    DT> Filter
    DT Filter> Clear Include
    DT Filter> Include TraceRulePerformance
    DT Filter> exit
  3. Archive some items into a classification enabled archive
  4. Stop the fsdmhost process by stopping the Storage service
  5. Open the DTrace log and notice that for each rule there is the percentage of time that was spent on it during classification

{RulesRunner.TraceRulePerformance} Average time to evaluate rules per item (ms) : 32
{RulesRunner.TraceRulePerformance} % Personal : 3.13549832026876
{RulesRunner.TraceRulePerformance} % Large number of attachments : 4.03135498320269
{RulesRunner.TraceRulePerformance} % Identity Card (Germany) : 2.23964165733483
{RulesRunner.TraceRulePerformance} % Low importance : 1.00783874580067
{RulesRunner.TraceRulePerformance} % Productivity documents : 7.61478163493841
{RulesRunner.TraceRulePerformance} % Web links : 2.68756998880179
{RulesRunner.TraceRulePerformance} % Auto-generated news feeds : 1.45576707726764
{RulesRunner.TraceRulePerformance} % Legal : 0.895856662933931
{RulesRunner.TraceRulePerformance} % Sensitive Project Code Names : 11.4221724524076
{RulesRunner.TraceRulePerformance} % Visa Card : 8.17469204927212
{RulesRunner.TraceRulePerformance} % Auto-reply : 1.00783874580067
{RulesRunner.TraceRulePerformance} % CPF Number (Brazil) : 6.83090705487122
{RulesRunner.TraceRulePerformance} % Email containers (attachments) : 0.559910414333707
{RulesRunner.TraceRulePerformance} % Discover Card : 2.46360582306831
{RulesRunner.TraceRulePerformance} % Message sent to external domain : 1.23180291153415
{RulesRunner.TraceRulePerformance} % Social Security Number (US) : 2.91153415453527
{RulesRunner.TraceRulePerformance} % American Express Card : 2.23964165733483
{RulesRunner.TraceRulePerformance} % Faxes (attachments) : 3.35946248600224
{RulesRunner.TraceRulePerformance} % Driving License (UK) : 1.23180291153415
{RulesRunner.TraceRulePerformance} % Company Confidential : 0.111982082866741
{RulesRunner.TraceRulePerformance} % Financial Data : 2.46360582306831
{RulesRunner.TraceRulePerformance} % Partial content : 1.45576707726764
{RulesRunner.TraceRulePerformance} % MasterCard : 3.58342665173572
{RulesRunner.TraceRulePerformance} % VAT/TVA number (France) : 3.2474804031355
{RulesRunner.TraceRulePerformance} % Large Items : 1.00783874580067
{RulesRunner.TraceRulePerformance} % National Insurance Number (UK) : 4.25531914893617
{RulesRunner.TraceRulePerformance} % Message sent to specific external domain : 1.3437849944009
{RulesRunner.TraceRulePerformance} % National Registry Identification Number (Singapore) : 0.335946248600224
{RulesRunner.TraceRulePerformance} % Current Retention Category Name : 4.14333706606943
{RulesRunner.TraceRulePerformance} % Charity solicitations : 7.27883538633819
{RulesRunner.TraceRulePerformance} % Permanent Account Number (India) : 6.27099664053751