Online communities are identified as people gathering online and communicating through the internet to share ideas, objectives, goals, without any geographical boundary. The growth of user-generated content created in online communities has transformed the way consumers search for and share information, particularly in the hospitality industry. Particularly, in the restaurant and food sectors due to the intangible nature of hospitality services, online reviews play an important role on consumer decisions. Furthermore, online reviews on restaurants are not only informational but also, they impact consumers’ choices regarding restaurants. Consequently, the nature of such user-generated content that is produced at a high speed and is diverse and rich should be treated and understood. This study proposes the first tailored BERTopic model together with sentiment analysis based on pre-trained BERT model that takes advantage of its novel sentence embedding for creating interpretable topics into the analysis of restaurant online reviews to determine how the customers elaborate their criteria in the context of certain experiences. An exploratory analysis is presented involving a large-scale review data set of 261,531 restaurant online reviews from 4 different countries retrieved from the eWOM community thefork.com. A broad list of the topics discussed by customers post-dining in restaurants is built. Insights into the behavior, experience, and satisfaction of the customers across the different restaurants are discovered. This approach and findings are encouraging hospitality managers in understanding customers’ perception, through which applicable marketing can be developed to attract and retain potential customers.

As Deepfakes phenomenon is spreading worldwide mainly through videos in web platforms and it is urgent to address the issue on time. More recently, researchers have extensively discussed deepfake video datasets. However, it has been pointed out that the existing Deepfake datasets do not properly reflect the potential threat and realism due to various limitations. Although there is a need for research that establishes an agreed-upon concept for high-quality datasets or suggests evaluation criterion, there are still handful studies which examined it to-date. Therefore, this study focused on the development of the evaluation criterion for the Deepfake video dataset. In this study, the fitness of the Deepfake dataset was presented and evaluation criterions were derived through the review of previous studies. AHP structuralization and analysis were performed to advance the evaluation criterion. The results showed that Facial Expression, Validation, and Data Characteristics are important determinants of data quality. This is interpreted as a result that reflects the importance of minimizing defects and presenting results based on scientific methods when evaluating quality. This study has implications in that it suggests the fitness and evaluation criterion of the Deepfake dataset. Since the evaluation criterion presented in this study was derived based on the items considered in previous studies, it is thought that all evaluation criterions will be effective for quality improvement. It is also expected to be used as criteria for selecting an appropriate deefake dataset or as a reference for designing a Deepfake data benchmark. This study could not apply the presented evaluation criterion to existing Deepfake datasets. In future research, the proposed evaluation criterion will be applied to existing datasets to evaluate the strengths and weaknesses of each dataset, and to consider what implications there will be when used in Deepfake research.

This paper first surveys three kinds of learner interview corpora (LINDSEI, NICT-JLE Corpus, and Trinity Lancaster Corpus), paying particular attention to their interview structures. Then, it explains the principles and features of the ICNALE Spoken Dialogue (ICNALE SD), which includes 425 videos and approximately 1.6-million-word transcripts of the L2 English interviews with 405 learners from ten regions in Asia and twenty native speakers. The ICNALE SD is one of the largest learner interview corpora and practically the sole dataset for the analysis of dialogue speeches by various Asian learners. As a case study based on the ICNALE SD, the author sought to find out how fluently learners in different regions speak in the interviews, which words they characteristically use, and which relationship is observed among them.

본 연구에서는 파랑수치모형(SWAN)을 사용하여 우리나라 연안역에서의 장기 파랑을 후측모델링하고, 그 활용성에 대하여 논하였다. 파랑후측모델링을 위한 입력 바람자료(NCEP, ECMWF, JMA-MSM)를 검토한 결과, JMA-MSM의 예측정확도가 높게 나타났지만 상대적으로 자료제공기간이 짧아 자료제공기간이 긴 ECMWF 바람자료를 채택하였다. 파랑후측모델링은 파랑관측부이가 설치되어 검증이 가능한 2001년부터 2014년까지의 ECMWF 바람자료를 이용하여 수행하였으며, 생성된 모델 결과는 기상청, 국립해양조사원의 파랑관측 부이자료를 이용하여 검증하였다. 파랑후측모델링 검증결과 파랑관측 부이자료와 잘 일치하였으며, 특히 태풍과 같은 이벤트기간의 해역 상황을 전반적으로 잘 재현하였다. 이를 통하여 현재 파랑관측부이 자료의 한계인 결측기간 동안의 파랑자료를 대체할 수 있음을 확인하였다. 하지만 일부 정점에서는 이벤트 기간 동안의 최대파고를 과소평가하는 것으로 나타났으며, 이러한 이유는 바람입력자료의 시간간격 및 해상도, 수심자료, 격자크기 등의 한계로 파악된다. 본 파랑후측모델링 결과는 연안역에서의 침식원인규명 특히, 이벤트 시기의 파랑특성과 연계한 분석이 가능하며, 원하는 연안지점에서의 파랑후측정보를 생산할 수 있어 연안재해취약성평가 등에 활용이 가능하다.

To understand the pattern recognition from dataset, a study should be started from the decomposition process of context into a collection of data pieces because the context may infer different words or information. Many researchers have been focused on finding an effective methodology for data storage, retrieval, representation, and discovery. As a similar endeavor, this paper proposes a new modeling method using group theory and situation theory. This paper provides how to construct a semi-group as a modeling method for pattern recognition from huge dataset. This process of construction of semi‐groups can be used as a retrieval tool for the decomposed information if necessary.

This paper is a study on data augmentation for small dataset by using deep learning. In case of training a deep learning model for recognition and classification of non-mainstream objects, there is a limit to obtaining a large amount of training data. Therefore, this paper proposes a data augmentation method using perspective transform and image synthesis. In addition, it is necessary to save the object area for all training data to detect the object area. Thus, we devised a way to augment the data and save object regions at the same time. To verify the performance of the augmented data using the proposed method, an experiment was conducted to compare classification accuracy with the augmented data by the traditional method, and transfer learning was used in model learning. As experimental results, the model trained using the proposed method showed higher accuracy than the model trained using the traditional method.

A robot usually adopts ANN (artificial neural network)-based object detection and instance segmentation algorithms to recognize objects but creating datasets for these algorithms requires high labeling costs because the dataset should be manually labeled. In order to lower the labeling cost, a new scheme is proposed that can automatically generate a training images and label them for specific objects. This scheme uses an instance segmentation algorithm trained to give the masks of unknown objects, so that they can be obtained in a simple environment. The RGB images of objects can be obtained by using these masks, and it is necessary to label the classes of objects through a human supervision. After obtaining object images, they are synthesized with various background images to create new images. Labeling the synthesized images is performed automatically using the masks and previously input object classes. In addition, human intervention is further reduced by using the robot arm to collect object images. The experiments show that the performance of instance segmentation trained through the proposed method is equivalent to that of the real dataset and that the time required to generate the dataset can be significantly reduced.

본 연구에서는 인공위성 및 재분석 자료인 Global Land Data Assimilation System (GLDAS), Global Land Evaporation Amsterdam Model (GLEAM), MOD16의 실제증발산량 산출물을 활용하여 한국수자원조사기술원(Korea Institute of Hydrological Survey, KIHS)에서 관리하고 있는 청미천(cheongmicheon farmland site, CFK)과 설마천(seolmacheon site, SMK) flux tower에서 검증하였고, Triple collocation (TC) 방 법을 활용하여 자료간의 불확실성 및 상관성분석을 수행하였다. 플럭스타워와의 검증 결과에서는 전반적으로 GLEAM>GLDAS>MOD16순으로 좋은 결과를 나타내었으며, 세가지 산출물의 조합(S1: flux tower vs. GLDAS vs. MOD16, S2: flux tower vs. GLDAS vs. GLEAM, S3: flux tower vs. GLEAM vs. MOD16)을 통한 TC 결과에서는 청미천(설마천)에서 GLEAM>GLDAS>MOD16>flux tower (GLDAS>GLEAM>MOD16>flux tower)순으로 좋은 결과를 나타내었다. TC 분석 결과에서 Flux tower의 error variance와 correlation coefficient가 상대적으로 좋은 결과를 나타내지 못하였으므로, 한반도 지역에서 인공위성과 재분석 자료(GLDAS vs. GLEAM vs. MOD16)만을 활용하여 TC를 적용하였다. 그 결과, GLDAS 와 GLEAM이 한반도 영역에서 낮은 error variance 와 높은 correlation coefficient를 나타낸 반면, MOD16의 경우, 농지에서 낮은 correlation coefficient과 높은 error variance를 나타내었다.

South Korea is a maritime nation, surrounded by water on three sides; hence, it is important to preserve in a sustainable manner. Most areas, especially those bordering the East Sea, have been suffering from severe coastal erosion. Information on the sediment yield of a river basin is an important requirement for water resources development and management. In Korea, data on suspended sediment yield are limited owing to a lack of logistic support for systematic sediment sampling activities. This paper presents an integrated approach to estimate the sediment yield for ungauged coastal basins by using a soil erosion model and a sediment delivery rate model in a geographic information system (GIS)-based platform. For applying the sediment yield model, a basin specific parameter was validated on the basis of field data, that, ranging from 0.6 to 1.2 for the 19 gauging stations. The calculated specific sediment yield ranged from 17 to 181 t/km2.yr in the various basin sizes of Korea. We obtained reasonable sediment yield values when comparing the measured data trends around the world with those in Korean basins.

This paper described the estimation of corn and soybeans yields of four states in the US Midwest using time-series satellite imagery and climate dataset between 2001 and 2012. We first constructed a database for (1) satellite imagery acquired from Terra MODIS (Moderate Resolution Imaging Spectroradiometer) including NDVI (Normalized Di°erence Vegetation Index), EVI (Enhanced Vegetation Index), LAI (Leaf Area Index), FPAR (Fraction of Photosynthetically Active Radiation), and GPP (Gross Primary Productivity), (2) climate dataset created by PRISM (Parameter-Elevation Regressions on Independent Slopes Model) such as precipitation and mean temperature, and (3) US yield statistics of corn and soybeans. ˜en we built OLS (Ordinary Least Squares) regression models for corn and soybeans yields between 2001 and 2010 after correlation analyses and multicollinearity tests. These regression models were used in estimating the yields of 2011 and 2012. Comparisons with the US yield statistics showed the RMSEs (Root Mean Squared Errors) of 0.892 ton/ha and 1.095 ton/ha for corn yields in 2011 and 2012 respectively, and those of 0.320 ton/ha and 0.391 ton/ha for soybeans yields. ˜is result can be thought of as a good agreement with the in-situ statistics because the RMSEs were approximately 10% of the usual yields: 9 ton/ha for corn and 3 ton/ha for soybeans. Our approach presented a possibility for extending to more advanced statistical modeling of crop yields using satellite imagery and climate dataset.

고해상도의 지형 데이터는 용량이 크기 때문에 GPU메모리에 데이터 전체를 적재할 수 없다. 따라서 out-of-core기반의 방법이 많이 사용된다. 그러나 보조기억장치의 대역폭 한계로 인하여 실시간으로 지형을 렌더링하기 어렵기 때문에 GPU로 웨이블릿 변환을 수행하여 압축된 DEM 데이터를 전송한 후 압축 해제하여 렌더링 하는 방법이 사용된다. 하지만 이 방법은 텍스처로부터 주기적으로 값을 읽어와 정점을 변환하고 메쉬를 생성해야하므로 비효율적이다. 이 논문에서는 웨이블릿 압축된 근사 계수 값을 정점의 속성으로 저장하고 기하 쉐이더에서 압축을 해제해 지형을 효율적으로 렌더링 하는 기법을 제안한다. 제안하는 방법은 근사 계수 값을 정점의 속성으로 주어 지형 텍스처의 전송량을 줄일 수 있다. 또한 지형 텍스처로부터 별도의 업로드 과정 없이 메쉬의 생성이 가능하므로 오버헤드가 발생하지 않아 효율적인 렌더링이 가능하다.

In an effort to examine the Regional Atmospheric Modeling System (RAMS ver. 4.3) to the initial meteorological input data, detailed observational data of NOAA satellite SST (Sea Surface Temperature) was employed. The NOAA satellite SST which is currently provided daily as a seven-day mean value with resolution of 0.1 o grid spacing was used instead of the climatologically derived monthly mean SST using in RAMS. In addition, the RAMS SST data must be changed new one because it was constructed in 1993. For more realistic initial meteorological fields, the NOAA satellite SST was incorporated into the RAMS-preprocess package named ISentropic ANalysis package (ISAN). When the NOAA SST data was imposed to the initial condition of prognostic RAMS model, the resultant performance of near surface atmospheric fields was discussed and compared with that of default option of SST. We got the good results that the new SST data was made in a standard RAMS format and showed the detailed variation of SST. As the modeling grid became smaller, the SST differences of the NOAA SST run and the RAMS SST43 (default) run in diurnal variation were very minor but this research can apply to further study for the realistic SST situation and the development in predicting regional atmospheric field which imply the regional circulation due to differential surface heating between sea and land or climatological phenomenon.