Retrieve New SD-Rules from OrgD, 2015
I. Description
A set of computer programs (FindSdRulesFromDPairs.java
) are developed to find the SD-Rules from a set of suffixD pairs. It identifies and eliminates the same starting characters of a SD-pair and then generates the SD-Rules automatically. Please notes that only root-parent SD-Rules is generated in this program. Two sets of SD-pairs are used for this task. This page details the new SD-Rules selected from orgD.
II. Procedures
Directory: ${SUFFIXD_DIR}
Programs:
shell>cd ${SUFFIXD_DIR}/bin
shell>GetSdRule ${YEAR}
2
orgFacts
shell>GetSdRule ${YEAR}
3
orgFacts
III. Results
- These are SD-Pairs from orgD of Lexicon.2013 (2014-15 has added new rules and thus eliminated some candidate rules)
- Once the new SD-Riles from orgD is added to the system, the suffixD form orgD will be decreased because they are generated in suffixD. So, 2013 release is used for this study.
- The file of ${ORG_D_DIR}/data/orgD.yes.S.data is used as input
- There are 4,110 SD-Pairs to generate 1,421 SD-Rules
- All generated SD-Rules are root-parent rules (without parent-rule).
- Rules with following criteria are selected:
- High frequency (>= 40)
- Accumulate coverage: 11.56% (> 11.50%)
- Individual coverage: 1.31% (> 1.00%)
- SD-Rules meet above criteria (total instance No. 4,110):
SD-RulesInstances No. | Accu. No. | Notes
| less$|adj|$|noun | 131 (3.19%) | 131 (3.19%) | Exists
| $|verb|ion$|noun | 111 (2.70%) | 242 (5.89%) | Exists
| ist$|noun|y$|noun | 63 (1.53%) | 305 (7.42%) | Exists
| ally$|adv|$|adj | 58 (1.41%) | 363 (8.83%) | New, with existing child rules
| ful$|adj|$|noun | 58 (1.41%) | 421 (10.24%) | Exists
| c$|adj|s$|noun | 54 (1.31%) | 475 (11.56%) | New, with existing child rules
| |
---|
- New SD-Rules with childred rules
Selected rules | Instances | Child-rules | Examples
|
---|
$|adj|ally$|adv | 58 |
ic$|adj|ically$|adv|2013|ORG_FACT|SELF
| - basic|adj|E0012047|basically|adv|E0218453
|
c$|adj|s$|noun | 54 |
ic$|adj|is$|noun|2013|ORG_FACT|SELF
| - gastritic|adj|E0029371|gastritis|noun|E0029372
|