• Korean Journal of Cognitive Science
• /
• v.4 no.2
• /
• pp.29-51
• /
• 1994

This paper presents a new approach to describing the meanings of English relative clauses with the theoretical framework of Conceptual Structure Theory (henceforth CST)which builds on the pionerring work of Sowa.And this paper aims at proposing some extensions to his work. CST describes the conceptual structrures of sentences with conceptual graphs(henceforth CG). which have begun to be used as an intermediate language in natural language processing and machine translation of computer.CGs are composed of concept types and conceptual relation types. They are a system of logic for semantic representation of sentences. This paper focuses on showing the differences of the CGs according to the functions of English relative clauses. English relative clauses are divided into restrictive and nonrestrictive uses.And this paper describes a restrictive clause with a CG including a expression.which derives from the viewpoint of Montague-semantics and Nom-S Analysis.This paper deals mainly with the relative clauses of double restroction as an example of restrictive relative clauses.The description of a nonrestrictive relative clause does not need any-expression, for it doesn's involve the meaning of set.And this paper links the CG of an appositive relative clause,which is a kind of nonrestrictive clauses,to the concept of the antecedent in the main clause.The description of a nonrestrictive relative clause with adverbial meaning is strated with two CGs for the main clause and the relative clause.They are linked with an appropriate intersentential conceptual relation type according to the contextual realtions between them.This paper also presents a CG of a sentential relative clause,which gives a comment on the main clause.

• Korean Journal of Cognitive Science
• /
• v.24 no.2
• /
• pp.95-109
• /
• 2013

With the rapid evolution of the personal device environment, the demand for natural language applications is increasing. This paper proposes a morpheme segmentation and part-of-speech tagging model, which provides the first step module of natural language processing for many languages; the model is designed for mobile devices with limited hardware resources. To reduce the number of morpheme candidates in morphological analysis, the proposed model uses a method that adds highly possible morpheme candidates to the original outputs of a conventional left-longest-match-preference method. To reduce the computational cost and memory usage, the proposed model uses a method that simplifies the process of calculating the observation probability of a word consisting of one or more morphemes in a conventional hidden Markov model.

• The Journal of the Korea Contents Association
• /
• v.13 no.6
• /
• pp.202-211
• /
• 2013

The research is based on the theory of cognitive psychology that a decision maker creates an internal belief through environment information process and that personal behavior is determined as a means of belief practice. This study implemented a language network analysis to understand the 16th President Noh Moo-hyun"s real estate policy belief who was a key decision maker of Korea. The results of this study are as follows. First, the chief variables of real estate policy recognized by president Rho Mu Hyun included civil economy and real estate speculation and so on. Second, the chief variables of real estate policy recognized by president Rho Mu Hyun did indicate weak correlation with other diverse economic variables. Third, the analysis shows that in practicing the real estate policy, excessive civil economy related with the real estate policy and the obsession with civil life variables worked as reason to pursue social contest rather than social agreement. It is thought that this study result will contribute to policy theory in explaining the policy decision-making process from the perspective of cognitive psychology.

• Korean Journal of Cognitive Science
• /
• v.22 no.1
• /
• pp.1-18
• /
• 2011

With rapid spread of handheld devices like cellular or smart phones, a short message service (SMS) comes on the public as a communication means. SMS is very cheap and can be easily written down on the storage in order not to forget it, hence it is widely used to inform schedules (time and place). In this paper, we develop a system for automatically registering schedules extracted from SMS text messages. SMS text messages are very short and concise, but include a lot of Internet words like slangs and abbreviations. These have made it difficult to extract information on schedules from them. Also handheld devices have some limitations on computing power and storage and then applying general natural language processing modules like morphological analysis to the devices are somewhat hard. To relax these burdens, we extract schedule informations from SMS messages using machine learning methods like condition random field (CRF) without using any language processing modules and register the informations on the schedule management system of handheld devices. Our proposed automatic schedule registration system has implemented on Samsung Omnia phone for experiments.

• Korean Journal of Cognitive Science
• /
• v.23 no.3
• /
• pp.295-322
• /
• 2012

In this paper, we cluster similar word senses applying vector space model and HAL (Hyperspace Analog to Language). HAL measures corelation among words through a certain size of context (Lund and Burgess 1996). The similarity measurement between a word pair is cosine similarity based on the vector space model, which reduces distortion of space between high frequency words and low frequency words (Salton et al. 1975, Widdows 2004). We use PCA (Principal Component Analysis) and SVD (Singular Value Decomposition) to reduce a large amount of dimensions caused by similarity matrix. For sense similarity clustering, we adopt supervised and non-supervised learning methods. For non-supervised method, we use clustering. For supervised method, we use SVM (Support Vector Machine), Naive Bayes Classifier, and Maximum Entropy Method.

• Korean Journal of Cognitive Science
• /
• v.24 no.1
• /
• pp.25-47
• /
• 2013

We conducted a psycholinguistic experiment in which participants orally produced sentences using a subject, a dative object, an accusative object, and a verb, provided just before the production. Results of the experiment are twofold: (i) Korean speakers basically produce the dative object earlier than the accusative one if the lengths of the objects are identical. (ii) If there is a length difference between the two objects, though, the longer one strongly tends to be placed before the shorter one, overriding the preference for 'dative-accusative' order. This 'long before short' preference which is generally observed in head-final languages appears to reflect the underlying tendency of the processing mechanism to put the heads of arguments and the predicate as closely as possible, thereby minimizing the cost for the processing of verb-argument structure.

• Korean Journal of Cognitive Science
• /
• v.16 no.2
• /
• pp.123-140
• /
• 2005

The purpose of this paper is to examine the frequency and usage of sentence patterns appearing in electronic dictionaries used in Korean language education in order to design an automatic sentence patterns checking. First, the concept of sentence patterns is defined and it is classified into sentence structure patterns and sentencial expression patterns. Sentence structure patterns and sentencial expression patterns are analyzed how they are expressed in the Korean Learner's Corpus. learner's Corpus is built into the Standard Corpus, which all Korean Learners must learn, and the Errors Corpus made by learners. From these research, we will find out how frequently the Sentential Patterns are being used in the Standard Corpus which has been made of Korean Texts and how the Sentential Pattern are being used in the Errors Corpus which were constructed from Korean learner's writings. Finally, having described the Sentential Patterns on the Sentential Electric Dictionary, we determine the optimum speed in the search for the Sentential Pattern.

• Korean Journal of Child Studies
• /
• v.9 no.2
• /
• pp.1-28
• /
• 1988

The purpose of this study was to develop a behavior rating scale for the evaluation of children's development for utilization by preschool teachers. The procedures for the study included content validation, pilot test, and main study. A total of 97 items were retained after the content validation and pilot test. The items of the scale were grouped into five areas (physical, language, cognitive, emotional, and social development) and 11 sub-areas. The resulting "Behavior Rating Scale for Preschool Children" was administered to 479 boys and girls, 3-6 through 6-5 years of age, selected from 10 different kindergartens and early education centers in Seoul, Pusan, and Chonju. The analysis of data was done with SPSS computer programs, including item analysis, Cronbach's ${\alpha}$ for reliability, factor analysis to test construct validity, two-way ANOVA to test age and sex differences, and percentile norms. The 97 items of the scale were found to be satisfactory in terms of item discrimination with indices ranging from .31 to .73. Cronbach's ${\alpha}$ was .98 for the total scale and ranged from .87 to .93 in specific domains, which was considered satisfactory. The factors extracted from each area were consistent with the educational objectives of the Yonsei Open Education Program except for emotional development. The intercorrelations among the domains were relatively high, ranging from .56 to .81. Age differences were significant in cognitive, physical, and language development, but not significant in social and emotional development. Sex differences were significant in all areas with girls higher on the average than the boys. Percentile ranks were drived from the total score for each age group and quartiles were calculated for sub-scores in each domain.

• Phonetics and Speech Sciences
• /
• v.15 no.1
• /
• pp.19-24
• /
• 2023

This study investigated acoustic (VOT) and articulatory characteristics of Korean velar stops in monosyllabic CV structures to examine how the three-way distinction is realized in the laryngeal and supralaryngeal domains and how the distinction is manifested in male versus female speakers' speech production. EMA data were collected from 22 speakers. In line with previous studies, male speakers preserved the three-way differentiation of velar stops (/k^*/</k/</k^h/) in terms of VOT while female speakers showed only a two-way distinction (/k^*/</k/=/k^h/). As for the kinematic characteristics, a clear three-way distinction was found only in male speakers' peak velocity measure in the C-to-V opening movement (/k^h/</k/</k^*/). For the other kinematic measures (i.e., articulatory closure duration, deceleration duration of the opening movement and the entire opening movement duration), male speakers showed only a two-way distinction between fortis and the other two stops. Female speakers did not show a three-way contrast in any kinematic measure. They showed a two-way distinction between lenis and the other two stops in C-to-V deceleration duration (/k^*/=/k^h/</k/), and a two-way distinction between fortis and lenis stops in the opening movement duration. An overall comparison of VOT and articulatory analyses revealed that the lenis-aspirated kinematic distinction is diminishing, driven by female speakers, in line with the loss of the lenis-aspirated distinction in VOT that could influence supralaryngeal articulation.

• Korean Journal of Cognitive Science
• /
• v.34 no.3
• /
• pp.197-226
• /
• 2023

Generative Text Summarization is one of the Natural Language Processing tasks. It generates a short abbreviated summary while preserving the content of the long text. ROUGE is a widely used lexical-overlap based metric for text summarization models in generative summarization benchmarks. Although it shows very high performance, the studies report that 30% of the generated summary and the text are still inconsistent. This paper proposes a methodology for evaluating the performance of the summary model without using the correct summary. AggreFACT is a human-annotated dataset that classifies the types of errors in neural text summarization models. Among all the test candidates, the two cases, generation summary, and when errors occurred throughout the summary showed the highest correlation results. We observed that the proposed evaluation score showed a high correlation with models finetuned with BART and PEGASUS, which is pretrained with a large-scale Transformer structure.