• Annual Conference on Human and Language Technology
• /
• 1999.10e
• /
• pp.462-469
• /
• 1999

본 연구에서는 시각적으로 제시되는 단어 자극의 의미 관련성의 차이에 따라 좌우반구의 처리가 어떻게 이루어지는지 알아보고자 하였다. 이를 위해서 명칭성 실어증 환자와 정상인 대학생 피험자를 대상으로 점화 어휘판단 과제를 수행하였다. 이 연구의 기본 논리는 명칭성 실어증 환자의 왼쪽 뇌가 손상되어 있기 때문에 어떤 정보처리가 왼쪽 뇌에서 일어나는 것이라면 정상인과 명칭성 실어증환자간의 수행에서 어떤 차이가 나타날 것을 기대되는 반면, 만일에 우뇌에서 처리되는 것이라면 정상인의 과제 수행 형태와 명칭성 실어증 환자의 것이 일치하는 형태를 보일 것이라는 것이다. 실험 1에서는 수직적 범주관련성이 어느 반구에서 정보처리 되는지를 조사하였다. 그 결과 정상인은 좌반구에서 유의미한 점화효과가 있고 우반구에서는 점화효과가 없었던 반면에, 명칭성 실어증 환자는 정상인과 정반대의 점화 효과를 보이고 있다. 이러한 결과는 좌반구가 일차적으로 수직적 범주 관련성 정보처리와 관련이 있음을 시사해 준다. 또한 수평적 범주 관련성에 따른 실험 은 정상인과 환자 두 집단 모두 수평적 범주관련성이 우반구에서 처리되는 유사한 패턴을 보여주었다. 실험2에서는 연합적 범주관련성에 따른 두 집단간의 점화 효과를 비교하였다. 정상인 집단과 환자 모두 좌우반구에 점화효과를 보여주고 있지만, 정상인 집단의 경우에는 우반구에서, 환자는 좌반구에서 점화량이 더 컸다. 연합관련 정보처리는 좌우반구 모두에서 일어난다고 하는 기존의 견해와 관련하여 볼 때 연합관련 정보처리는 좌우반구에서 일어난다고 해석할 수 있을 것이다. 명칭 실어증 환자의 정보처리는 정상인과 다르게 이루어지므로 이러한 좌우반구에서의 차이가 난 것으로 볼 수 있다. 이상의 실험1과 2의 결과를 종합해 보면, 시각적으로 제시되는 단어의 범주적 관련성이 주는 어휘정보 처리는 반구에 따라 처리하는 기능이 다르다고 결론 내릴 수 있다. 즉, 좌반구는 수직적 범주 관련성을 담당하고 우반구는 수평적 관련성을 담당하며, 연합적 관련성은 좌우반구 모두에서 정보처리 된다는 것이다.

• Korean Journal of Cognitive Science
• /
• v.31 no.3
• /
• pp.77-102
• /
• 2020

Some progress has been made in the study of the category coherence effect, which states that the attributes of soldiers or nuns with similarities in dress and behavior, and easily distinguished from other categories, are likely to be generalized. However, few studies have examined the fundamental psychological mechanisms that underlie this category coherence effect, and this study aims to fill this gap. For this purpose, two experiments were conducted after selecting categories with high coherence (nuns, soldiers, and flight attendants) and those with low coherence (interpreters, wedding planners, and florists). In experiment 1, we observed that the members of a category were presumed to have certain reasons to prefer [property X] (presumption of reasons for preference), with this presumption becoming stronger when [property X] was observed repeatedly in high-coherence categories than in the case of low-coherence categories. Experiment 2 showed that for the high-coherence categories, the presumption of reasons for preference was stronger when [property X], rarely seen in everyday life (base rate of 30%), was observed, while the presumption of reasons for preference was weaker when [property Y] (base rate 70%), frequently seen in everyday life, was observed. In the low-coherence categories, the presumption of reasons for preference tended to be weak for both rare and frequent attributes. That is, there were significant effects of the two-way interaction between category coherence and base rate on the presumption of reasons for preference. This study has implications for psychological essentialism and stereotyping.

• Science of Emotion and Sensibility
• /
• v.10 no.3
• /
• pp.491-502
• /
• 2007

Categorization is very important for product design. Consumer's emotion become different according to a type of categorization, so design concept and design elements must be combined differently with difference of the emotion. Specially, categorization process is necessary if nowadays product line is enlarged, and a product differentiation is not clear. That is, designers decide on correct categories and a design concept based on similarity of emotion and have to provide to consumer-oriented design. The purpose of this study is to provide a design direction for Korean traditional sauce products after extracting consumers' sensitivity from the whole image of Korean traditional sauce and each images of the sauces-korean hot pepper paste, soybean paste, fermented soybeans paste, SsamJang, and soy sauce- and deciding categories of the each sauces based on the extracted sensitivities' similarity. In the result of this study, we knew that Korean traditional sauces didn't differentiate from consumers' preference images. In our empirical research, the research - emotional image survey on sauces - have conclusion that emotional image of "well-being", "tasty" have positive influence, but emotional image of "messy and dirty", "smelly" have negative influence. Therefore, we suggest that positive emotional images like "tasty" should be emphasized, but negative emotional images like "messy" should be eliminated for design and marketing strategy of Korean traditional sauces. This research will suggest the guideline for product design with respect to academic aspects and working-level aspects.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.10
• /
• pp.487-492
• /
• 2017

Clustering is a technique which is used to measure similarities between data in big data analysis and data mining field. Among various clustering methods, k-Modes algorithm is representatively used for categorical data. To increase the performance of iterative-centric tasks such as k-Modes, a distributed and concurrent framework Spark has been received great attention recently because it overcomes the limitation of Hadoop. Spark provides an environment that can process large amount of data in main memory using the concept of abstract objects called RDD. Spark provides Mllib, a dedicated library for machine learning, but Mllib only includes k-means that can process only continuous data, so there is a limitation that categorical data processing is impossible. In this paper, we design RDD for k-Modes algorithm for categorical data clustering in spark environment and implement an algorithm that can operate effectively. Experiments show that the proposed algorithm increases linearly in the spark environment.

• Science of Emotion and Sensibility
• /
• v.17 no.1
• /
• pp.39-52
• /
• 2014

Use of vision and audition for video reality has made much advancement. However use of olfaction, which is effective in inducing emotion, has not yet been realized due to technical limitations and lack of basic research. In particular it is difficult to fabricate many odors required for each different video. One way to resolve this is to discover clusters of odors of similar smell and to use representative odor for each cluster. This research explored clusters of odors based on pairwise similarity ratings. 300 diverse odors were first collected and sorted them into 11 categories. We selected 152 odors based on their frequency, preference, and concreteness. Participants rated similarity on 1,018 pairs of odors from selected odors and the results were analyzed using multi-dimensional scaling (MDS). Based on the idea that low odor concreteness would support valid use of representative odor, the MDS results are presented from low to high smell concreteness. First, flowers, plants, fruits, and vegetables was classified under the easy categories to use representative odor due to their low smell concreteness (Figure 1). Second, chemicals, personal cares, physiological odors, and ordinary places was classified under the careful categories of using it due to their intermediate concreteness (Figure 2). Finally, food ingredients, beverages, and foods was classified under the difficult categories to use it because of their high concreteness (Figure 3). The results of this research will contribute to reduction of cost and time in odor production and provision of realistic media service to customers at reasonable price.

• Journal of Korean Library and Information Science Society
• /
• v.47 no.2
• /
• pp.51-70
• /
• 2016

The purpose of this study is to analyze the semantic relation network and valence-arousal dimension through the words that describe emotions in Korean language. The results of this analysis are summarized as follows. Firstly, each emotion word was semantically linked in the network. This particular feature hinders differentiating various types of "emotion words" in accordance with similarity in meaning. Instead, central emotion words playing a central role in a network was identified. Secondly, many words are classified as two categories at the valence and arousal level: (1) negative of valence and high of arousal, (2) negative of valence and middle of arousal. This aspects of Korean emotional words would be useful to analyze emotions in various text data of books and document information.

• The Journal of Society for e-Business Studies
• /
• v.25 no.3
• /
• pp.77-93
• /
• 2020

The literature has reported that hierarchical classification methods generally outperform the flat classification methods for a multi-class document classification problem. Unlike the literature that has constructed a class hierarchy, this paper evaluates the performance of hierarchical and flat classification methods under a situation where the class hierarchy is predefined. We conducted numerical evaluations for two data sets; research papers on climate change adaptation technologies in water sector and 20NewsGroup open data set. The evaluation results show that the hierarchical classification method outperforms the flat classification methods under a certain condition, which differs from the literature. The performance of hierarchical classification method over flat classification method depends on class similarities at levels in the class structure. More importantly, the hierarchical classification method works better when the upper level similarity is less that the lower level similarity.

• The Journal of the Korea Contents Association
• /
• v.13 no.1
• /
• pp.19-29
• /
• 2013

In the current Internet environment where there is high space complexity of information, search engines aim to provide accurate information that users want. But content-based method adopted by most of search engines cannot be used as an effective tool in the current Internet environment. As content-based method gives different weights to each web page using morphological characteristics of vocabulary, the method has its drawbacks of not being effective in distinguishing each web page. To resolve this problem and provide useful information to the users, this paper proposes an evaluation method based on categories. Category-based evaluation method is to extend query to semantic relations and measure the similarity to web pages. In applying weighting to web pages, category-based evaluation method utilizes user response to web page retrieval and categories of query and thus better distinguish web pages. The method proposed in this paper has the advantage of being able to effectively provide the information users want through search engines and the utility of category-based evaluation technique has been confirmed through various experiments.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.79-79
• /
• 2017

강우 자료는 수공구조물 설계목적에 따라 다양한 시공간적 범주가 필요하다. 그러나 시간단위 이하 시계열 강우자료는 미계측 유역 및 관측연한 등의 제약으로 연속적인 시계열을 확보하는데 어려움이 있다. 이러한 점에서 포아송분포 기반 강우발생모형은 강우시계열의 통계적 특성을 나타내는 5개 매개변수로 다양한 시간 범주의 연속강우시계열을 생성할 수 있다는 장점이 있다. 강우발생모의 핵심은 과거자료의 통계특성을 효과적으로 복원할 수 있어야 하며, 다양한 기상학적 특성들 또한 적절하게 모의될 수 있어야 한다는 점이다. 즉, 다음과 같은 기준으로 모의적합성을 평가할 수 있다. 첫째, 지속기간별 관측시계열과 모의시계열의 통계적 유사성을 평가하고, 둘째, 확률분포를 따르는 각 매개변수의 사후분포를 제시하여 불확실성을 정량화하고, 셋째, 추정된 매개변수의 물리적 범위의 적정성 검토가 필요하다. 본 연구에서는 강우발생모형으로 널리 알려진 Neyman-Scott Rectangular Pulse(NSRP) 모형과 Bayesian 모형을 연계한 Bayesian NSRP 모형 개발을 통해 강우관측소 전지점에 대한 매개변수 지도를 제시하고자 한다. 본 연구결과는 임의 유역에 대한 강우발생 시나리오를 제공하여, 다양한 형태의 유출결과를 도출할 수 있으며, 무엇보다 유출결과를 확률적으로 평가할 수 있다는 장점이 있다.

• Journal of KIISE:Databases
• /
• v.37 no.2
• /
• pp.71-81
• /
• 2010

The problem of finding clusters is widely used in numerous applications, such as pattern recognition, image analysis, market analysis. The important factors that decide cluster quality are the similarity measure and the number of attributes. Similarity measures should be defined with respect to the data types. Existing similarity measures are well applicable to numerical attribute values. However, those measures do not work well when the data is described by categorical attributes, that is, when no inherent similarity measure between values. In high dimensional spaces, conventional clustering algorithms tend to break down because of sparsity of data points. To overcome this difficulty, a subspace clustering approach has been proposed. It is based on the observation that different clusters may exist in different subspaces. In this paper, we propose a new similarity measure for clustering of high dimensional categorical data. The measure is defined based on the fact that a good clustering is one where each cluster should have certain information that can distinguish it with other clusters. We also try to capture on the attribute dependencies. This study is meaningful because there has been no method to use both of them. Experimental results on real datasets show clusters obtained by our proposed similarity measure are good enough with respect to clustering accuracy.