PD Source Model - Lexicon Suffix Derivations with Negations

I. Introduction

The negation rule can be applied in the derivation morphology to retrieve antonyms [2006 Paradis]. Prefix derivations with negation tags in the Lexicon are retrieved as aPairs candidates. For example, negative prefixes of “un-“, “in-“, “non-“ and “count-“ for aPairs of [available|unavailable|adj], [correct|incorrect|adj], [diabetic|nondiabetic|noun], and [clockwise|counterclockwise|adv], respectively.

II. Design

Antonym candidates are retrieved from the Lexicon prefixD (derivation.data.${YEAR}). These antonym candidates are retrieved from the following prefixes: a-, an-, anti-, contra-, counter-, de-, dis- ,dys-, il-, im-, in-, ir, mis-, non-, un-, under- with five POS (noun, verb, adj, adv and prep).

It is possible that prefixes in a prefix derivation pair (dPair) are not included in the above list, yet, it is a valid antonym. For example, extra- is the prefix of ordinary|extraordinary, which is a legit prefix and aPair. Also, antonym pairs might include multiple prefix derivations, such as overdose|underdose which involves prefix derivations of dose|underdose and dose|overdose. The above two cases of aPairs from prefix derivations can be found in the model of co-occurrence in a corpus [CC] and thus are not discussed here.

Please see design documents for more details.

III. Implementation

Java source codes are implemented in the directory of Derivation:

• GetAntCandFromPrefixD.java

Input derivations:

Use the prefixD in the Lexicon as source. The prefix derivations have 9 fields:

Algorithm:

• if Type = prefixD (P)
• if Negation = negative (N)
• if derivation-1 and derivation-2 meet criteria of legal antonyms
• if same POS

• get derivations of Derivation-1
• can't be the same as Derivation-2
• convert to citation form and output to aPair candidate format

Output APairs: