• Journal of KIISE:Software and Applications
• /
• v.31 no.12
• /
• pp.1649-1659
• /
• 2004

Text Generation is a process of generating comprehensible texts in human languages from some underlying non-linguistic representation of information. Among several sub-processes for text generation to generate coherent texts, this paper concerns referring expression generation which produces different types of expressions to refer to previously-mentioned things in a discourse. Specifically, we focus on pronominalization by zero pronouns which frequently occur in Korean. To build a generation model of referring expressions for Korean, several features are identified based on grammatical information and cost-based centering model, which are applied to various machine learning techniques. We demonstrate that our proposed features are well defined to explain pronominalization, especially pronominalization by zero pronouns in Korean, through 95 texts from three genres - Descriptive texts, News, and Short Aesop's Fables. We also show that our model significantly outperforms previous ones with a 99.9% confidence level by a T-test.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.9
• /
• pp.379-390
• /
• 2019

In this paper, we propose a novel deep neural network model for Vision-and-Language Navigation (VLN) named LVLN (Landmark-based VLN). In addition to both visual features extracted from input images and linguistic features extracted from the natural language instructions, this model makes use of information about places and landmark objects detected from images. The model also applies a context-based attention mechanism in order to associate each entity mentioned in the instruction, the corresponding region of interest (ROI) in the image, and the corresponding place and landmark object detected from the image with each other. Moreover, in order to improve the success rate of arriving the target goal, the model adopts a progress monitor module for checking substantial approach to the target goal. Conducting experiments with the Matterport3D simulator and the Room-to-Room (R2R) benchmark dataset, we demonstrate high performance of the proposed model.

• Journal of the Korean Institute of Educational Facilities
• /
• v.28 no.4
• /
• pp.11-17
• /
• 2021

This study is aimed at presenting an efficient management plan of user participatory design in a situation where the School Space Innovation Project is in progress. 2 schools that were the targets of the Incheon School Space Innovation Project in 2019 were selected for this, and features such as the physical environment of that classroom, classroom usage plan, and the stepwise outcome of the user participatory design workshop were contemplated. Especially the workshop outcome was compared and analyzed quantitatively, focusing on the actual master plan and zoning plan, in order to identify the feature that opinions of various users are reflected on the actual plan. As a result, the following conclusion could be reached. Firstly, it was confirmed that the expression about the user preferential space influences the classroom usage plan of that classroom. Vague expressions about the whole space held a large majority of the objects for the linguistic expression of the preferential space. The expression mode as limited as the expression of the actions that users want to carry out in the space. On the other hand, when the usage purpose of the classroom was definite, it was confirmed that the demand for furniture·facility is relatively high. Secondly, according to the analysis of zoning for each function, it seems that the stereotype, which is arranged on the basis of the chalkboard at the front of existing classrooms, was applied in the case of the learning zone. However, in cases of other functions, a tendency was identified that the user carries out an image description that reflects the physical features of the space. Sufficient preparation will need to precede for the efficient management of the user participatory design workshop and the acceptance of various opinions. It seems that especially the classroom usage plan, number of workshops, consultation of each step, and the education about the space expression mode affect the master plan.

• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.191-204
• /
• 2021

Visual Question Answering (VQA) and image captioning are tasks that require understanding of the features of images and linguistic features of text. Therefore, co-attention may be the key to both tasks, which can connect image and text. In this paper, we propose a model to achieve high performance for VQA by image caption generated using a pretrained standard transformer model based on MSCOCO dataset. Captions unrelated to the question can rather interfere with answering, so some captions similar to the question were selected to use based on a similarity to the question. In addition, stopwords in the caption could not affect or interfere with answering, so the experiment was conducted after removing stopwords. Experiments were conducted on VQA-v2 data to compare the proposed model with the deep modular co-attention network (MCAN) model, which showed good performance by using co-attention between images and text. As a result, the proposed model outperformed the MCAN model.

• Journal of the Korea Computer Graphics Society
• /
• v.2 no.1
• /
• pp.39-47
• /
• 1996

Consumer's needs, pursuing diverse lifestyles, can be identified systematically by the development of design research method with computer technologies. Color, which is the most important factor in industrial design, has been regarded as possessing difficulties in collecting and analyzing reliable data, because it has multi-dimensional features. The purpose of this paper is to develop a new research method for design and color preference investigation, and to provide the possibilities of applying this method into the traditional color research which has many limitations in time, space, and money. This paper emphasizes that the new method using visual and concrete 3D modeling of a product can enhance the reliability of collecting and analyzing data comparing with traditional linguistic and abstract one.

• Korean Journal of Human Ecology
• /
• v.19 no.1
• /
• pp.63-74
• /
• 2010

The purpose of the study was to examine the actual conditions of caregiver-infant ratios, group-room activity areas, evaluations of infant programs and caregiver-infant interactions based on structural and process indicators which are major factors of infant care. The subjects were 20 caregivers and 91 infants from 14 infant classes of 13 day care centers in Daejeon. An actual survey was conducted on caregiver-infant ratios and group-room activity areas, and teaching-learning plans for infants and daily schedules were gathered for the evaluation of infant programs. The caregiver-infant interactions were observed every one minute for a total of 20 minutes using Lee Wan Jeong's "Evaluation Measure of Caregiver-infant Interactions"(1999). The results of this study were as follows: First, caregiver-infant ratios ranged from 2.5 to 7 infants per caregiver, resulting in the difference of the number of infants. Second, the 14 classes for one-year-old infants were arranged in three different ways; 5 classrooms with distinctive activity areas, 2 without any divided areas and 7 containing a mix of partial activity areas. Third, in teaching-learning plans for infants, there were a large number of topics related to seasonal features and experiences while the fewest were about basic life habits. Fourth, in the caregiver-infant interactions, caregivers used more positive interactions and linguistic modeling than sensitive responses to infants and social interactions.

• Journal of Digital Convergence
• /
• v.10 no.1
• /
• pp.365-369
• /
• 2012

We propose a vision-based method that estimates train position and movement for railway monitoring in which we use a fuzzy classifier to determine train states. The proposed method employs frame difference and background subtraction for estimating train motion and presence, respectively. These features are used as the linguistic variables of the fuzzy classifier. Experimental results show that the proposed method can correctly estimate train position and movement. Therefore the method can be used for railway monitoring systems which estimate crowd density or protect safety.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2005.05a
• /
• pp.23-27
• /
• 2005

Humans perceptually segment events, but models that predict where events will be segmented are limited. Developing a detailed model may be hard because of the overlapping quality of events (i.e., one can smile and walk at the same time, but the endpoint of each event can be different). However, some aspects of events appear to be universally represented in the world's languages. For example, path, the trajectory of an object's movement, is one of the most universally encoded event features. Although it is generally encoded in the prepositions of English (e.g., up), in other languagesit is encoded in the verbs (e.g., descendere). Linguistic universals may represent basic levels of event perception. Here we consider how one of these, path, might be parsed. Because the spatiotemporal projection of paths to an observation point is similar to the spatial projection of objects, we tested the hypothesis that path segmentation and object segmentation would be based on similar image properties, such as discontinuities in orientation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.4
• /
• pp.598-603
• /
• 2003

This paper presents a sub-band feature extraction approach in which the feature extraction method in the individual frequency sub-bands is determined in terms of speech recognition accuracy. As in the multi-band paradigm, features are extracted independently in frequency sub-regions of the speech signal. Since the spectral shape is well structured in the low frequency region, the all pole model is effective for feature extraction. But, in the high frequency region, the nonparametric transform, discrete cosine transform is effective for the extraction of cepstrum. Using the sub-band specific feature extraction method, the linguistic information in the individual frequency sub-bands can be extracted effectively for automatic speech recognition. The validity of the proposed method is shown by comparing the results of speech recognition experiments for our method with those obtained using a full-band feature extraction method.

• The Journal of the Korea Contents Association
• /
• v.4 no.4
• /
• pp.133-139
• /
• 2004

This paper proposes a Korean sentence boundary detection system which employs k-nearest neighbor algorithm. We proposed three scoring functions to classify sentence boundary and performed comparative analysis. We uses domain independent linguistic features in order to make a general and robust system. The proposed system was trained and evaluated on the two kinds of corpus; ETRI corpus and KAIST corpus. As experimental results, the proposed system shows about $98.82\%$ precision and $99.09\%$ recall rate even though it was trained on relatively small corpus.