개요

• 중의성을 해소하기 위한 WSD approach 중 Selectional preferences(선택 선호도) 개념을 평가하는 기법으로 Pseudo-words를 제안한 논문

  While pseudo-words are now less often used for word sense disambigation, they are a common way to evaluate selectional preferences, models that measure the strength of association between a predicate and its argument filler, e.g., that the noun lunch is a likely object of eat. Selectional preferences are useful for NLP tasks such as parsing and semantic role labeling(Zapirain et al., 2009). Since evaluating them in isolation is difficult without labeled data, pseudoword evaluations can be an attractive evaluation framework.

• backoff model을 제안함

  • Katz의 백오프 모델이란 ?

• PDF (+ 필기)

Pseudo-Words

• 유사 어휘/ 유사 비단어(類似非單語,pseudoword)란 어떤 특정언어의 음운규칙에 맞고 그 언어에 속한 단어처럼 보이지만 실제로는 존재하지 않는 단어를 말한다.

• WSD(Word sense disambiguation, 중의성 해소 문제)를 해결하기 위한 방법론

  One way to mitigate this problem is with pseudowords, a method for automatically creating test corpora without human labeling, originally proposed for word sense disambiguation (Gale et al.,1992; Schutze, 1992).

• the concatenation of origin word/confounder word

  A pseudo-word is the concatenation of two words (e.g. house/car). One word is the original in a document, and the second is the confounder. Consider the following example of applying pseudo-words to the selectional restrictions of the verb focus:

  Original: This *story focuses on the campaign.* Test: This *story/part focuses on the campaign/meeting.*

  In the original sentence, focus has two arguments: a subject story and an object campaign. In the test sentence, each argument of the verb is replaced by pseudo-words. A model is evaluated by its success at determining which of the two arguments is the original word.

기존 평가법의 문제 2가지

• Selectional preferences는 subsets of data(unseen words 또는 특정 빈도 범주 안의 단어)에 포커싱

  →모든 all entire test examples에 대한 평가법을 제안

• confounder를 어떻게 선택하는지가 다양 (최적의 방법도 아니며, task difficulty)

  → random confounders 사용 & nearest-neighbor frequencies 사용 (오버피팅까지 방지)

성과

we present a surprisingly simple baseline that outperforms the state-of-the-art and is far less memory and computationally intensive. It outperforms current similarity-based approaches by over 13% when the test set includes all of the data. We conclude with a suggested backoff model based on this baseline.