• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.222-227
• /
• 2023

환각 현상이란 LLM이 생성 태스크에서 사실이 아닌 내용을 생성하거나 근거가 없는 내용을 생성하는 현상을 말한다. 환각 현상은 LLM이 생성한 출력물에 대한 사용자의 신뢰를 떨어뜨리기 때문에 환각을 완화할 수 있는 방법이 필요하다. 최근 사후 편집 모델 중 하나인 RARR는 입력 텍스트를 질문들 순서에 따라 순차적으로 편집하여 환각을 완화하였지만 이전 단계의 편집 오류가 전파되거나 같은 작업을 반복하는 등의 단점이 있었다. 본 논문은 환각 현상 완화를 위한 단위 사실 기반 사후 교정을 제안한다. 제안한 방법은 입력 텍스트를 단위 사실로 분해하고 각 사실에 대응하는 질문을 생성한 후 검색된 관련 문서로 환각 여부를 판단한다. 환각이라 판단되면 편집을 수행하여 환각을 완화한다. 병렬적으로 편집을 진행하기 때문에 기존 연구의 순차적인 오류 전파 문제를 해결하고 기존 연구에 비해 더 빠른 사후 편집을 진행할 수 있다. 실험 결과, 제안 방법이 RARR보다 Preservation Score, 원문과의 사실성 일치여부, 의도 보존 여부에서 모두 우수한 성능을 보인다.

• Journal of Digital Contents Society
• /
• v.11 no.2
• /
• pp.177-184
• /
• 2010

These days, as digital producing technique has been developed, 3D imaging technique is used in high-tech computer and T.V. Also study for 3D producing technique is actively in progress. Moreover, as James Cameron's movie, 'Avatar' released in 2009 was a box office hit, the issue about 3D image came to the fore again. At this point, I decided to study the effect of the visual depth caused by the color correction during the post-production stage. The purpose of this study is to offer information about processing effective images through data about the effect of the visual depth that applies the color correction during the post-production stage. Basically, I supposed that color and contract would have effects on depth of 3D image. As a result, I could find out the changes of visual depth, space perception and sense of depth throughout the experiment. Applying this result,, I produced the 15 minutes of 3D advertisement movie and I found out that the color correction during the post-production stage was very effective for 3D depth. The left image and the right image by beam splitter based rig and parallel rig were used for this study. Also I adjusted the strong contrast by the color correction during the post-production stage after correcting convergence and visual depth during editing. As a result, I could produce images which had strong sense of space and sense of depth.

• Cartoon and Animation Studies
• /
• s.40
• /
• pp.139-155
• /
• 2015

Animation is completed in accordance with the original planning of visual direction in the process of post-production. Especially, a variety of visualization skills to meet the planning idea are directed in editing and compositing processes such as rendering and shading in the production of 3D animation. Therefore, the studies how to implement these visual effects have been conducted since the emergence of the animation. The rendering method which appears in the recent animations so often is the effect of 2D animation style although it is applied by 3D technique. 3D animation can be classified as two types including photo realistic and non-photo realistic renderings. The former is to render the images realistically while the latter is to highlight the formative effect with analogue style free from the realism of the photographs. Visual effect of 2D animation is non-realistic, that is, non-photo realistic rendering. The animations produced by this method have the depth of 3D animation expression in terms of space and animating, and can direct the viewer-friendly analogue style visuals, which are widely used as a post-production effect. Hence in this study, we selected a short animation, 'Paperman,' one of Disney's animations, produced by cartoon rendering method which belongs to the class of non-photo rendering technique in 3D graphics, as our subject. We analyzed their techniques and visual effects of the scenes expressed by cartoon rendering method so as to understand whether they meet the original direction idea with the composition work. In addition, we expect further developments of post-production methods, exceeding the limit of graphic expression in alignment with the trend that has become more various in the types and genres of non-photo rendering.

• The Journal of the Korea Contents Association
• /
• v.9 no.1
• /
• pp.205-214
• /
• 2009

There has been a change in the production of the movie and techniques of the expression in the digital technology which has been used universally since 2000. As the focus of the production system of the movie has moved from the production procedure centering around filming to the procedure centering around editing used by the digital tools. The flows and boundaries of the production system which is consist of pre-production, production and post-production are getting indistinct. In the point of view of aesthetics and technology, this kind of change expands the creativity and originality for the screenwriters and provides the reality and vitality which haven't been found in previous films for the audience. It also elevates the quality of the movie. However, this ‘illusionary spectacle' has a vulnerable point that shows the aura and details of the movie soundly for the audience.

• Journal of Digital Convergence
• /
• v.9 no.1
• /
• pp.89-106
• /
• 2011

With the development of the Internet, the media market is experiencing rapid change. The traditional form of offline newspaper markets, which has created the mainstream news delivery market, have been losing large amounts of power due to the emergence of online media. On the other hand, online media armed with its ability to post breaking news almost as it occurs has increased its influence and popularity at a rapid speed. However, since the exposure time and readership of online newspapers are less than their offline counterparts online media has a high dependence on sensational titles and somewhat exaggerating headlines rather than the objective and recapitulative headlines which are preferred. This paper will examine the function and significance of headlines, and conduct a comparative analysis on contents and forms of the headlines in offline newspaper and online newspaper in order to identify the characteristics of each newspaper. Through comparisons, it examines the types of headlines and the transference of headlines. In addition, it examines the differences in recognition of copy editors according to media in determining the headline, and it illuminates which type of headlines is preferred in order to instigate readers to read online newspapers. The analysis results of title-transferring cases in offline newspaper and online newspapers are that offline newspapers try to convey the contents accurately by making use of an information-transferring headline, while attention-attracting headline or subject-hiding headlines tend to be used for attracting the eye of readers. In addition, analysis results of headlines that are put on the top with a click counts in online newspapers are that attention-attracting headlines are the majority, while subject-hiding headlines that are hardly used in offline newspapers, are well-used in online newspapers. In an interview with copy editors, they responded that offline editors focus on the fact-transference in determining the title, while online editors put the weight on immediacy in the delivery of news and the function of guiding readers.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.49 no.2
• /
• pp.13-19
• /
• 2012

The techniques of 3d image processing have broadly used in the areas including movies, games, performances, exhibitions. In addition, increasing demands for practical uses have gradually extended to the areas of architecture, medicine, nuclear power plant. However, dominant techniques for 3d image processing seem to depend on multi-camera in which two stereo images are merged into one image. Also the pipeline has limitations to provide real-time 3d viewer in ubiquitous computing. It is not able to be applicable onto most general screens as well. In addition, the techniques can be utilized for the real-time 3d game play without a particular monitor or convertor. Hence, the research presented here is to aim at developing an efficient real-time 3d viewer using only mono camera which do not need post processing for editing as well.

• Geophysics and Geophysical Exploration
• /
• v.4 no.4
• /
• pp.133-144
• /
• 2001

An effective seismic reflection technique for mapping the cavities and bedrock surface in carbonate rocks is described. The high resolution seismic reflection images were successfully registered by using the hydrophones employed in the stream-water driven trench, and were effectively focused by applying optimal data processing sequences. The strategy included enhancement of the signal interfered with the large-amplitude scattering noise, through pre- and post stack processing such as time-variant filtering, bad-trace editing, residual statics, velocity analysis, and careful muting after NMO (normal moveout) correction. The major reflections including the bedrock surface were mapped with the desired resolution and were correlated to the seismic crosshole tomographic data. Shallow major reflectors could be identified and analyzed on the AGC (auto gain control)-applied field records. Three subhorizontal layers were identified with their distinct velocities; overburden (<3000 m/s), sediments (3000-4000 m/s), limestone bedrock (>4000 m/s). Taking into account of no diffraction effects in the field records, gravel-rich overburdens and sediments are considered to be well sorted. Based on the images mapped consistently on the whole survey line and seismic velocity increasing with depth, this area probably lacks in sizable cavities (if any, no air-filled cavities).

• Korean Journal of Poultry Science
• /
• v.48 no.2
• /
• pp.81-90
• /
• 2021

Avian influenza viruses (AIVs) must utilize host cellular factors to complete their life cycle, and fragile X mental retardation protein (FMRP) has been reported to be a host factor promoting AIV ribonucleoprotein (vRNP) assembly and exports vRNP from the nucleus to the cytoplasm. The functional role of chicken FMRP translational regulator 1 (cFMR1) as a host factor of AIV is, however, poorly understood. In this study, we targeted the cFMR1 gene in DF1 cells using clustered regularly interspaced short palindromic repeats/Cas9-mediated genome editing to examine the functional role of cFMR1 as a host factor of AIV. We found that cFMR1 stimulated viral gene transcription during early stages of the viruses' life cycle and did not affect viral progeny production and viral polymerase activity in DF1 cells 24 hours post infection. cFMR1 overexpression did not exert significant effects on virus production, compared to the control. Therefore, unlike in mammalian systems (e.g., humans or mice), cFMR1 did not play a pivotal role in AIV but only seemed to stimulate viral proliferation during early stages of the viral life cycle. These results imply that the interplay between host factors and AIV differs between mammals and avian species, and such differences should be considered when developing anti-viral drugs for birds or establishing AIV-resistant bird models.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.1
• /
• pp.239-244
• /
• 2022

Neural Machine Translation (NMT) is gradually breaking down the boundary between human and machine translation. We look at actual cases of human and machine translation and discuss why machine translation needs a human touch. In this paper, we raise three driving questions: Can humans be replaced by machines?; How human translators can remain successful in a NMT-driven world?; Is it possible to eliminate language barrier in the era of NMT and World Englishes? The answers to these questions are all negative. We suggest that machine translation is a useful tool with rapidity, accuracy, and low cost productivity. However, the machine translation is limited in the areas of culture, borrowing, ambiguity, new words and (national) dialects. The machines cannot imitate the emotional and intellectual abilities of human translators since machines are based on machine learning, while humans are on intuition. The machine translation will be a useful tool that does not cause moral problems when using methods such as back translation and human post-editing. To conclude, we propose the blended approach that machine translation cannot be completed without the touch of human translation.

• Journal of Korean Neurosurgical Society
• /
• v.65 no.2
• /
• pp.204-214
• /
• 2022

Objective : Osteoporosis result from age-related decline in the number of osteoblast progenitors in the bone marrow. Probiotics have beneficial effects on the host, when administered in appropriate amounts. This study investigated the effects of probiotics expressing specific genes, especially the effects of genetically modified bone morphogenetic protein (BMP)-2-expressing Lactobacillus plantarum CJNU 3003 (LP) on ovariectomized rats. Methods : Twenty-eight female Wistar rats (250-300 g, 12 weeks old) were divided into four groups : the sham (control), the ovariectomy (OVX)-induced osteoporosis group (OVX), the OVX and LP (OVX/LP), OVX and genetically modified BMP-2-expressing LP (OVX/LP with BMP) groups. The three groups underwent bilateral OVX and two of these groups were administered two different types of LP via oral gavage daily. At 16 weeks post-OVX, blood was collected from the heart and the bilateral tibiae were extracted and were scanned by ex-vivo micro-computed tomography and stained with hematoxylin-and-eosin (H&E) and Masson's trichrome stain for pathological assessment. The serum levels of osteocalcin (OC), rat C-telopeptide of type I collagen (CTX-I), BMP-2, and receptor activator of nuclear factor-ĸB ligand (RANKL) were measured. Results : The 3D-micro-computed tomography images showed that the trabecular structure in the OVX/LP with BMP group was maintained compared with OVX and OVX/LP groups. No significant differences were detected in trabecular thickness (Tb.Th) between control and OVX/LP with BMP groups (p>0.05). Furthermore, a tendency toward increased BMD, trabecular bone volume, Tb.Th, and trabecular number and decreased trabecular separation was found in rats in the OVX/LP with BMP groups when compared with the OVX and OVX/LP groups (p>0.05). The H&E and Masson's trichrome stained sections showed a thicker trabecular bone in the OVX/LP with BMP group compared with the OVX and OVX/LP groups. There was no difference in serum levels of OC, CTX and RANKL control and OVX/LP with BMP groups (p>0.05). In contrast, significant differences were found in OC and CTX-1 levels between the OVX and OVX/LP with BMP groups (p<0.05). Conclusion : Our results showed that the expression of genetically modified BMP-2 showed inhibition effect for bone loss in a rat model of osteoporosis.