• Journal of KIISE
• /
• v.44 no.2
• /
• pp.179-185
• /
• 2017

Document summarization aims to generate a summary that is consistent and contains the highly related sentences in a document. In this study, we implemented for document summarization that extracts highly related sentences from a whole document by considering both similarities and entailment relations between sentences. Accordingly, we proposed a new algorithm, TextRank-NLI, which combines a Recurrent Neural Network based Natural Language Inference model and a Graph-based ranking algorithm used in single document extraction-based summarization task. In order to evaluate the performance of the new algorithm, we conducted experiments using the same datasets as used in TextRank algorithm. The results indicated that TextRank-NLI showed 2.3% improvement in performance, as compared to TextRank.

• English Language & Literature Teaching
• /
• v.9 no.2
• /
• pp.173-192
• /
• 2003

The purpose of this paper is to analyze the semantic ambiguity of coordination connected by 'and' in English and show a clear explanation for the meaning of the sentence. Two sentences which are connected by 'and' can be divided into the symmetric conjunction and the asymmetric conjunction according to its meaning. Especially, in order to explain the different meaning in the asymmetric conjunctions, we introduce the theory of presupposition and entailment, which provides the solution to the ambiguity in meaning. It can be said that presupposition and entailment can logically explain the meaning of natural language.

• Language and Information
• /
• v.7 no.2
• /
• pp.163-178
• /
• 2003

Korean displays an interesting construction (so-called possessor agreement construction), where a possessor nominal and its possessum nominal are marked with the same case as shown in the example Mary-ka John-ul tali-lul cha-ss-ta ‘Mary kicked John's leg’ More interestingly, not all possessors in possessive construction are marked the same case with its possessum as shorn in the ungrammatical sentence ＊Mary-ka John-ul cha-lul cha-ss-ta ‘Mary kicked John's car’. Hence, a simple but non-trivial question arises: In what situation are both possessors and possessums marked with the same case\ulcorner In this paper, we advance three claims: (i) Possessor agreement appears in the situation where entailment is satisfied as follows: If Mary kicked John's leg, it entails that Mary kicked John, (ii) entailment in possessor agreement results from theta-feature sharing; specifically, the whole DP and the possessor DP share the same theta role, and (iii) Possessor nominals are marked with accusative (or nominative) case when they are assigned internal theta role from the predicate directly.

• Language and Information
• /
• v.18 no.2
• /
• pp.123-140
• /
• 2014

The aim of this paper is twofold: to identify two types of even if concessive conditionals, standing-ifs and introduced-ifs (Bennett 1982) in terms of whether the truth of the consequent is 'entailed' (the consequent-entailment problem in terms of Lycan 2001); and to analyze these two types of concessive conditionals in a compositional way. Here we argue that, following Guerzoni and Lim (2007), even if conditionals can be analyzed as the cases where even gets its focus in conditionals. We also argue that the consequent-entailment problem can be accounted for in a compositional way if we identify the focus as well as the scope of even in conditionals correctly. We further argue that the analysis presented in this paper supports the scope theory of even, among two theories of even previously proposed in various works. We also consider the possibility where concessive conditionals without even can be analyzed as an extension of the analysis proposed in this paper.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2007.11a
• /
• pp.115-125
• /
• 2007

The purpose of this paper is to propose a new type of NPI licensing context through French subjunctive and ne expletif. The distribution of NPIs on previous studies does not exactly correspond to negative function types. French subjunctive and ne expletif are good guidelines for reclassifying NPI licensing context. My classification is by a hierarchy of strength in negative force: overtly negative proposition > negative entailment > negative implicature. A new type of NPI licensing context is: (i) I-domain for negative implicature (ⅱ) E-domain for negative entailment and (ⅲ) overt negation.

• Language and Information
• /
• v.4 no.1
• /
• pp.9-15
• /
• 2000

• Proceedings of the Korean Society for Language and Information Conference
• /
• 2000.06a
• /
• pp.9-15
• /
• 2000

• Language and Information
• /
• v.15 no.1
• /
• pp.79-92
• /
• 2011

Some terms related to "meaning" in semantics are reviewed, particularly those used in Lyons' famous semantics book Semantics 1 and Semantics 2 (Cambridge University Press, 1977). Since Lyons' book is so comprehensive that the semantic terms used in that book cover almost every important notions/terms used in contemporary semantics. With relevant interpretation of the terms, we suggest most appropriate Korean translations, trying to give different translations for different English terms. Terms covered include: 'signify, signification; reference/refer/referent, sense; denotation/denote, connotation/connote; extension, intension; implication/imply, implicature/implicate, entailment/entail; designate, demonstrate/demonstrative'.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.377-381
• /
• 2023

재순위화기 연구는 주로 파이프라인 과정 설계, 데이터 증강, 학습 함수 개선, 혹은 대규모 언어 모델의 지식 활용 등에 집중되어있다. 이러한 연구들은 좋은 성능 상승을 이끌어주었지만 실제 적용이 힘들 뿐만 아니라 학습 비용이 크게 발생한다는 한계점을 가지고 있다. 더 나아가 주어진 데이터 집합만을 활용해서는 보다 더 세부적인 학습 신호를 주기 어렵다는 단점 또한 존재한다. 최근 자연어처리 분야의 연구에서는 피드백을 인위적으로 생성하여 반영하여 모델 성능 상승을 이끄는 연구가 제안되었다. 본 연구는, 이러한 연구를 바탕으로 질의와 문서 간의 함의 관계 점수를 피드백으로 사용 및 재순위화기 모델로의 반영을 제안한다. 재순위화기 모델에 대해 피드백을 반영하는것은 그렇지 않은 모델 대비하여 성능 상승을 이끌며 피드백 반영이 더 좋은 표상 도출에 도움이 됨을 확인할 수 있다.

• Korean Journal of Logic
• /
• v.19 no.2
• /
• pp.145-184
• /
• 2016

Williamson(1994) proved incompatibility of Gab Theory and Tarski T-schema. But this does not means that Gab Theory could not involve intuition on truth that is expressed by T-schema. I will show that Gab Theory and mutual entailment of 'p' and 'it is true that p'(p⊨T

and T

⊨p) are compatible. It will draw that Gab Theory can involve minimal intuition on truth. After all what I want to reveal is logical space for Gab Theory through the compatibility of the mutual entailment and negation of the Principle of Bivalence. To prove the compatibility, I will present a consequent relation which should be accepted whenever we accept Gab Theory and demonstrate Gab Theory and the mutual entailment imply following two thesis; 1) not-T

and T are not equivalent. 2) p entails T

but not-T

does not entails not-p.