Example - Add SD-Rules from Other Suggestions

Other suggested SD-Rules can be evaluated and added to this system by following the same approach in the previous two sessions: nomD session and factD session.

• Suggested SD-rules:

  Suggested SD-Rule	Notes	Examples	Status
  $|verb|less$|adj	This rule should be evaluated first to complete suffix negation.	count$|verb|countless$|adj resist$|verb|resistless$|adj tame$|verb|tameless$|adj tire$|verb|tireless$|adj	Evaluated: Bad SD-Rule
  $|adj|ly$|adv	Suggested by Lynn	quick|adj|quickly|adv	Rejected - duplicated
  $|noun|y$|adj	Suggested by Lynn	wood|noun|wody|adj	Rejected - duplicated
  al$|adj|us$|noun	Suggested by Chris, seconded by Destinee & Lynn	viral|adj|virus|noun estral|adj|estrus|noun coccal|adj|coccus|noun	TBD
  $|noun|ize$|verb	Suggested by Lynn	terror|noun|terrorize|verb	TBD
  e$|verb|ing$|noun	Suggested by Chris	rave|verb|raving|noun	TBD

• Apply the same procedures to get the optimized set by using the latest optimized set of 2.3.4.5 as new baseline. This task involves:
  • Retrieve all raw SD-pairs from Lexicon (2013) of above four selected SD-rules
  • Tag raw SD-pairs
  • Get stats of SD-pairs of these four SD-rules
  • Add to SD-rules set and find the optimization
    • Check if root parent-rule
  • The total valid SD-Pair no. (TotalYes) needs to be calculated as total valid SD-pair no. from all parent-rules.

  The iterative results are shown as follows:

The result shows the new selected SD-rule does not improve the system performance and should not be included in the optimized set. However, this SD-rule is added to the SD-rules evaluation set to cover more evaluated SD-rules. The new optimized set has coverage rate (95.01%) and system performance (1.9031) with accuracy rate of 95.30% to include 73 (out of 96) SD-rules. This set should be used as baseline for evaluating new rules in the future. The diagram below shows the system accuracy and coverage curves of this optimized set.