조사시점에 따른 반복질문이 피조사자의 심리상태에 미치는 영향

백주미 경기대학교 대학원 2018 국내석사

본 연구는 조사상황에서 사용되어지는 암시적 질문유형의 하나인 반복질문의 실제적 효과성을 확인하기 위해 참여자들의 사고 및 감정을 중심으로 사건발생 이후 조사시점에 따른 반복질문의 암시성을 확인하고자 하였다. 질문의 방식은 첫 질문에 대한 참여자의 응답 직후 선행된 질문과 동일한 두 번째, 세 번째, 네 번째 질문을 반복질문으로 하여 ‘첫 질문을 하였을 때’와 ‘반복질문을 하는 과정에서’로 나누어 참여자들의 사고 및 감정을 비교분석하였다. 사건발생 이후 각 시점에 무선할당된 참여자들은 ‘직후’ 집단 22명, ‘일주일 후’ 집단 22명, ‘이주일 후’ 집단 21명으로 총 65명의 자료가 분석에 사용되었다. 연구결과, ‘사고’는 각 집단내에서 전체적으로 유의한 차이가 있는 것으로 나타났다. 평균 50% 정도의 참여자가 본 실험에서 제한하였던 금지된 행동에 대해 사실과는 무관하게 자신의 잘못으로 내재화하였으며, 불안, 두려움, 죄책감, 걱정 등의 감정을 함께 경험한 것으로 나타났다. 그러나 ‘감정’은 ‘이주일 후’ 집단을 제외한 두 집단에서 유의한 차이가 발생하지 않았다. 각 조사시점이 다른 세 집단간의 비교에서는 ‘첫 질문을 하였을 때’, ‘사고’는 차이가 있는 것으로 나타났으나 ‘감정’의 차이는 발생하지 않았고, ‘반복질문을 하는 과정에서’의 세 집단간의 ‘사고’와 ‘감정’의 차이는 발생하지 않았으며 허위자백율의 차이도 발생하지 않았다. 이는 시간적 요인에 관계없이 반복질문의 암시성이 사람들의 기억 및 신념을 바꾸고 이와 동반된 감정들도 자백하기 이전 범죄인에게서 나타난 감정들과 다르지 않았다는 것을 보여준 결과로서, 조사자의 고암시적 질문방식은 사람들로 하여금 내재화의 가능성을 높이고 무엇이 옳고 그른지 알지 못하여 문제의식 조차 가지기 어렵게 할 수 있다. 따라서 조사상황에서 이러한 질문방식의 사용은 자제하여야 할 것이다. This study was to investigate the implication of thoughts and emotion of the examinees statements about repeated questions according to investigative time point in order to determine the actual effectiveness of repeated questions, a suggestive question type. The thoughts and emotions of the examinees measured at first question and repeated questions were analyzed and compared with second, third, fourth questions identical to previous ones just after the examinees response to the first question. The data collected from 65 subjects assigned randomly, after the occurrence of event, into 'just after' (n=22), 'after one week, (n=22), and 'after two week' (n=21) groups were used in analysis. The results showed that the 'thought' was significantly different within each group. Approximately half of subjects were shown to internalize that they committed the forbidden act regardless of the actual fact and experience, in addition, emotions such as anxiety, fear, guilty, and concern. The 'emotion', however, was observed to be significantly different only within 'after two weeks' group. Three groups with different investigative time point showed that, in the first question, the differences among group were observed in the 'thought' but not in the 'emotion' and in the repeated questions, there were no differences in both the 'thought' and 'emotion' among groups nor in the false confession rate. These results showed that the implication of the repeated questions alters people's memory and belief and that the accompanied emotions are not different from those observed in criminals bofore they confess, suggesting that the question containing strong implication is likely to increase the internalization of the examinees and to make their judgement on right and wrong and deprive them of even critical mind. It is considered, therefore, that such questioning method is to be avoided.

몸의 경험으로서의 미래 : 딥러닝 모델을 이용한 실시간 미래 영상 생성과 작품화

백주미 서강대학교 영상대학원 2022 국내박사

본 연구는 주로 과거와 현재로 국한되었던 인터랙티브 미디어아트에서의 시간 경험을 미래의 축으로 확장하고자 하는 문제 제기로부터 출발한다. 우리의 시간 경험은 이미 일어난 일에 대한 경험뿐 아니라 아직 일어나지 않은 미래에 대한 경험도 포함한다. 아직 경험하지 못한 일에 대한 기대, 예측 자체가 우리가 가지는 미래의 경험이다. 그리고 이는 우리 몸의 다양한 시간 메커니즘의 작동에 기반한다. 이러한 맥락에서 미래는 우리 몸의 경험으로 이해할 수 있다. 우리가 의식하든 못하든 세계에서 가지는 대부분의 경험에는 미래의 경험이 수반된다. 그런데 시간성을 주제로 다루었던 그간의 인터랙티브 작업은 미래의 경험을 다루지 못했다. 이는 사용 가능한 영상 소스가 현재와 과거의 영상으로 제한되었기 때문이다. 본 연구의 목적은 실시간으로 미래를 예측할 수 있는 진보된 기술을 연구하고 이를 예술적으로 활용하여 현재와 과거의 시간에 제한되었던 예술 경험을 미래로 확장하는 데에 있다. 이를 위해 실시간으로 관객의 다음 동작을 예측하는 인공신경망 기술을 연구하고 이를 활용한 인터랙티브 미디어아트를 제안하였다. 시퀀스 데이터를 다루는 RNN(Recurrent Neural Network)을 이용하였으며 두 개의 RNN 모듈로 이루어진 sequence-to-sequence 아키텍쳐를 사용하였다. 실시간 인터랙티브 미디어아트에서 관객이 자신의 시간과 작품 속 시간을 동시에 일어나는 것으로 느끼는 동시성과 경험이 끊기지 않고 계속되게 하는 지속성을 중요한 지향점으로 삼았다. 데이터셋 구축, 모델 학습부터 작품 구현을 위한 알고리즘 설계까지 전반적인 과정에서 시간 경험의 동시성과 지속성을 유지하기 위한 방법을 연구하고 실험하였다. 관객의 가능 미래를 잘 예측하기 위해서는 데이터셋도 새로 만들 필요가 있었다. 전시장에서 관객이 할만한 행동들을 예상하여 800여 개의 포즈 시퀀스로 된 Gallery Action 데이터셋을 만들었다. 실시간 인터랙티브 아트에서 지속성과 동시성을 가지기 위해 데이터셋의 시퀀스 길이를 길게 하여 랜덤한 구간을 학습할 수 있도록 하였다. 최종적으로 seq2seq 모델을 사용하여 관객의 실시간 가능 미래를 예측하는 세 점의 인터랙티브 작품을 제안하였다. 각 작품은 각각 다른 시간의 양상들을 모티브로 한다. 첫 번째 작품 <몸이 되어버린 시간으로서의 나>는 항상 과거와 미래로 향하는 현재 시간의 지향성을 모티브로 하여 항상 변화하려는 주체의 지향성을 표현하였다. 두 번째 작품 <Stochastic Present>와 세 번째 작품 <Stochastic World>는 매순간 다양한 미래가 생성된다는 양자물리학의 다세계 해석을 모티브로 삼는다. 우리의 현재가 다양한 가능 미래를 만들어 내는 모습을 직접 경험할 수 있도록 하였다. 각 작품마다 다른 시간의 양상들을 모티브로 하고 있으나 시간을 몸과의 관계를 통해 체험하도록 한 것은 세 작품 모두의 공통적 바탕이다. 본 연구는 인공지능 기술의 예술적 활용을 통해 현재와 과거의 시간에 제한되었던 관객의 경험을 미래까지 확장해주었다. 이는 더 다양한 시간의 예술적 표현과 경험을 가능하게 해준다. 인간의 감각 운동 능력으로 가질 수 없는 시간 경험은 존재와 세계를 이해하는 데에 새로운 시각을 제시할 수 있을 것이다. 과거-현재-미래를 포괄하는 체화된 시간 경험은 모든 종류의 경험과 상호작용의 기반이 되기 때문에 본 연구에서 제안하는 미래 예측 모델은 상호작용적 경험이 중요한 다양한 분야에도 넓게 활용될 수 있을 것으로 기대한다. This study starts from the need to expand the time experience in interactive media art, which was mainly limited to the past and present, to the axis of the future. Our experience of time includes not only experiences of what has already happened, but also experiences of the future that have not yet happened. Expectations and predictions of things we have not yet experienced are our future experiences. And it is based on the operation of various time mechanisms in our body. In this context, the future can be understood as the experience of our body. Every experience in the world is always accompanied by an experience of the future. However, the interactive media art with temporality as the subject did not deal with future experiences. This is because the available video sources of viewers were limited to current and past images. The purpose of this study is to study advanced technology that can predict the future in real time and to use it artistically to expand the artistic experience limited to the present and past time into the future. To this end, we researched artificial neural network technology that predicts the next action of the audience in real time, and proposed interactive media art using it. A sequence-to-sequence architecture consisting of two RNN modules was used. In this study, the simultaneity and continuity of time experience in real-time interactive media art were taken as important directions. Simultaneity means that the audience feels their own time and the time in the work as occurring at the same time, while continuity means that the experience continues without interruption. We researched and experimented with methods to maintain the simultaneity and continuity of temporal experience in the overall process from data set construction and model learning to algorithm design for realization of the work. In order to predict the possible future of the audience well, it was also necessary to create a new dataset. In anticipation of the actions that the audience would perform in the exhibition hall, we created a Gallery Action dataset. In order to have continuity and simultaneity in real-time interactive art, the length of data sequence is lengthened so that random sections can be learned throughout the repeated learning process. Finally, three interactive art works using the seq2seq model to predict the possible future of viewers were proposed. Each work is based on different aspects of time. The first work <I as time that has become a body> expresses the tendency of the subject to always change. This coincides with the tendency of the present time to move toward the past and the future. The second work <Stochastic Present> and the third work <Stochastic World> take as their motif the multi-world interpretation of quantum physics that claim various futures are created every moment. It allows viewers to experience themselves how our present creates a variety of possible futures. Each work has different aspects of time as its motif, but the common ground of all three works is that time is experienced through the relationship with the body. This study extended the audience's experience, which was limited in the present and past, to the future through the artistic application of artificial intelligence technology. This enables artistic expression and experiences of a wider variety of times. Since the embodied temporal experience encompassing the past-present-future is the basis for all kinds of experiences and interactions, the future prediction model proposed in this study is expected to be widely used in various fields where interactive experiences are important.

상호문화주의적 관점에서 본 피나바우쉬(Pina Bausch)의 안무경향에 관한 연구

백주미 경희대학교 대학원 2012 국내석사

The purpose of this study is to analyze the inter-culturism shown through the analysis of Pina Bausch's dance pieces that are having significant effects in the advancement of today's modern dance circle in order to search for various and practical possibilities of inter-culturism in dance, as well as where cultures can coexist in harmony in the global village era of the 21st century. Accordingly, the inter-culturism tendency shown in her dance pieces was analyzed according to the theme, movement, music and stage composition through which the artistic values inter-culturism in contemporary dance choreography are presented as follow. First, combining with other cultures and art genres will ensure variety of choreography. Second, promoting exchange between cultures will allow the choreography that strive towards internationalization. Third, inter-culturism choreography tendency can induce open communication between cultures. It is hoped that this study can provide opportunity for expanding the area of dance by showing variety even in the dance theme and movement upon the advancement into deeper art through the inter-culturism tendency among the various elements of contemporary dance.

노인 건강증진을 위한 창의적 움직임 콘텐츠 개발과 U-헬스케어 모델 연구

백주미 경희대학교 대학원 2017 국내박사

In this age, the aged are suffering from depression, dementia, or emotional instability. Such problems of the aged are not just personal problems but becoming social issues. Therefore, it is needed to develop programs that can influence mental and emotional health positively in the aged. Appropriate physical activity in old age is one of the essential elements to maintain physical and mental health and lead a successful life. Either dance movement or physical expression activity can be regarded appropriate as a program to solve such problems mentioned above. Creative movement contents equipped with effects like creative body expression based on sensory stimuli, improvement in immersion and concentration with diversified experiences, and also mutual communication and social exchange can be properly used as a program to enhance physical, mental, and emotional health in the aged. Furthermore, along with the dramatic development of ICT, the fourth industrial revolution is being accelerated, and health care systems are also being converted into ubiquitous health care (U-health care) or such. Now is the time to develop, extend, and apply dance contents aggressively. We should study and develop dance contents like creative movement contents so that they can be the core products of ICT-based health improving service industry and distribute them to the U-health care market. As part of it, this author is going to suggest dance contents that can influence dance contents’ popularization and commercialization positively and reduce financial or medical cost in the country as they can be used as a preventive senior health improvement program with contents provided as an online app. through cable-media or wearable devices. The purpose of this study is to find out ways to improve health in the aged in Korea currently revealing aging phenomena dramatically. This researcher will develop, apply, and verify creative movement contents and suggest them as a model applicable to the ICT-based U-health care system. The results of this study are as written below: First, considering the characteristics of the aged, this study has developed 『The Awakening of the Body and Emotion』, an educational and participative dance program with creative movement contents supporting their health and leisure.『The Awakening of the Body and Emotion』 is a program with creative movement contents developed to enhance health in the aged, and the physical movement contents have been reorganized through literature review as well as a Delphi survey to experts. These contents are intended for a broader meaning of movement than dance, and they are focused on inducing body movement or motional expressions by applying a variety of creative methods and means. It has three steps: opening the senses, training, and growing. With ‘imagining’ as the basic element of it, it is indeed to help the aged to recover their physical and mental senses and grow. 12 topics based on that are classified by establishing motivations for learning, Body, Shape, Effort, and Space (BESS movement). In each session, there is a step for ‘imitating’, and it is designed and developed to help them understand the class’ topic and principle and then move their body freely as they wish. Second, this author has examined the senior health improving dance program, 『The Awakening of the Body and Emotion』, with creative movement contents positively and verified if it is effective in relieving their depressive symptoms. To 12 aged participants, a survey was conducted total three times (before the sessions begin, after six sessions, & right after the completion of 12 sessions) by applying “Geriatric Depression Scale-Short Form: Senile Depression”. According to the results, 11 out of the total 12 participants indicated improvement in depression, which means that creative movement contents are effective in relieving or improving depression in the aged. Aside from such positive results, through the group interview conducted in open space before and after each session, various opinions were collected, for example, ‘I am satisfied about recovering the forgotten senses of my body.’, ‘I was glad that I could experience a new sort of thought or imagination.’, ‘I felt revitalized physically.’, or ‘I got familiar with other participants as time went by.’ Like that,『The Awakening of the Body and Emotion』 with creative movement contents has been proven to be positively effective in relieving aged participants’ depression and influence their physical․mental․social health positively based on their opinions. Third, based on the positive results, to popularize and industrialize 『The Awakening of the Body and Emotion』 with creative movement contents, this author suggests an ICT-based U-health care model. 『The Awakening of the Body and Emotion』 with creative movement contents consists of service products to improve physical․mental․social health in the aged and provides new services affiliated with an ICT-based U-health care service provider. Applying the contents to an ICT-based U-health care service system, this researcher suggested it to Application, Cable-Media, Smart Device, sanatoriums/hospitals for the aged, and welfare organs for the aged as a service model in order to popularize and industrialize the contents. To increase the applicability of them as a service model, this author has surveyed and interviewed the participants about whether they intend to use the dance class of 『The Awakening of the Body and Emotion』 when it is provided as a service and gained following results about evaluation on it. Female participants choose ‘Yes’ or ‘Medium’ while male participants respond ‘Yes’ or ‘Absolutely yes’, and they say they intend to use the contents out of curiosity or interest in improving their health by the service. The results of the survey and interview support the applicability of 『The Awakening of the Body and Emotion』 with creative movement contents as a new model of U-health care. The present government in 2017 takes its senior welfare policy mainly to support dementia in the aged, improve their health, and increase their pension. The government’s coping with it and multilateral political measures imply they have much concern about aging society and aging-related problems. Along with that, it is needed to recognize the functions and roles of dance newly to maintain and improve health in the aged, develop many programs, and provide them as a model to allow the aged to lead a healthy life in senescence. Also, ICT-based dance contents and creative movement contents can be used to solve either socioeconomic or cultural anthropologic problems and enhance health in the aged. Furthermore, they will turn out to be the good cases which let each individual aged person to realize their own hopes and dreams to ‘improve and actualize life quality’ and ‘lead a happy and satisfying life in senescence’.

Energy management strategies for university campus buildings using a data-driven energy mapping method

백주미 Graduate School, Yonsei University 2022 국내박사

In Korea, excluding apartments for residential use, hospitals (11.8%), offices (10.5%), and universities (10.2%) consume the most energy. Yonsei University ranks second in greenhouse gas emissions among the universities in Korea. In terms of energy consumption in energy intensive buildings, electricity is the highest at 79.3% regardless of the size of energy consumption. University campuses comprise buildings of various uses and types, which implies that energy management of university campus buildings should be approached differently from that of individual buildings. In this study, by applying the data-driven energy mapping method to analyze the energy usage pattern of campus buildings and suggesting a prediction method using the energy map, a method for energy management of the entire university campus has been suggested. This will help reduce greenhouse gas emissions at universities. The Yonsei University campus was chosen as the subject of this study, which includes a total of 63 buildings, and there are 27 measurement points that collect electricity consumption; hourly electricity consumption was used, which excluded monthly gas consumption. The energy map expresses energy usage data for 365 days a year and 8760 hours in a heat map format. Based on the electric energy consumption per unit area, it consists of 10 steps and checks the energy usage by color. One block represents one hour, and the first row represents a day from 00:00 to 23:00 h and consists of 365 columns. The energy mapping process involves collecting data, missing data imputation, building an energy map to form a base model, predicting energy consumption, updating the base model, and applying the predictive model to the operation. In the energy mapping method, a machine learning algorithm was applied, and the Gradient Boosted Trees (GBT) algorithm was applied to a model for missing data imputation, an energy consumption prediction model, and a model for operation. Each prediction model satisfies ASHRAE Guideline 14, CVRMSE 30%, which is the acceptance criterion for time unit prediction. By using the energy mapping method developed in this study, the energy map of each building can be grouped by usage pattern to understand the consumption characteristics of individual buildings. Additionally, the energy mapping of the entire campus allows us to check the peak demand prediction, rendering it easy to manage through the energy usage map for each building at each hour. In this study, an algorithm for determining the optimal start/stop time was presented to suggest the possibility of energy saving. Application of the algorithm revealed that the heating period of the Yonsei-Samsung Library can be used for at most 5 to 1 h per floor or date. Improvements were made to reduce the operation time, which is equivalent to as much as 32.27% and as little as 5.6% of the total operating time. Key words: Energy Mapping Method, Energy Map, University Campus Energy Management, Gradient Boosted Trees, Electricity Consumption Prediction 전국을 기준으로 주택용도인 아파트를 제외하고 병원(11.8%), 오피스(10.5%), 대학(10.2%) 순으로 에너지 사용량이 많다. 연세대학교는 전국에 있는 대학교 중에서 온실가스 배출량 2 위를 차지하고 있다. 에너지 다소비 건물의 에너지원 사용 현황은 에너지사용량 규모와 상관없이 전기가 79.3%로 사용비중이 가장 높다. 대학교 캠퍼스는 다양한 용도와 유형의 건물로 구성되어 있다. 이는 대학교 캠퍼스 건물의 에너지 관리는 단순히 개별건물의 에너지관리와 다르게 접근해야 한다는 것을 알 수 있다. 본 연구에서는 데이터를 기반으로 하는 에너지 맵핑 기법을 적용하여 캠퍼스의 에너지 사용량 패턴을 분석하고 에너지맵을 이용한 예측방법을 제시함으로써 대학교 캠퍼스 전체 에너지 관리에 활용할 수 있는 방안을 제시하고자 한다. 이는 대학교 온실가스 감축에도 도움이 되고자 한다. 본 연구의 대상지는 연세대학교 캠퍼스로 총 63 개의 건물을 포함하고 있으며 전력사용량을 수집하는 계측포인트는 27 개이다. 본 연구에서는 월별로 수집되는 가스사용량은 제외하고 1 시간 단위로 계측되는 전력사용량을 사용하여 진행하였다. 에너지맵이란 1 년 365 일, 8760 시간의 에너지 사용량 데이터를 히트맵 형식으로 표현한 것이다. 단위면적당 전력사용량을 기준으로 10 단계로 구성하여 컬러로 에너지 사용량을 확인한다. 1 개의 블록이 1 시간을 나타내며 1 행은 0 시부터 23 시까지의 하루를 나타내며 365 개의 열로 구성되어 있다. 에너지 매핑 기법은 데이터의 수집, 결측치 대체, 베이스 모델이 되는 에너지맵 구축, 에너지 사용량 예측, 베이스모델 업데이트, 예측 모델을 운영에 적용하는 프로세스를 거친다. 에너지 매핑 기법 프로세스는 머신러닝 알고리즘을 적용하였는데, 결측치 대체를 위한 모델, 에너지 사용량 예측 모델, 운영에 적용하기 위한 모델에 Gradient Boosted Trees(GBT) 알고리즘을 적용하였다. 각 예측모델은 ASHRAE Guideline 14, 시간단위 예측의 허용 기준인 CVRMSE 30%를 만족하였다. 본 연구의 결과물인 에너지 매핑 기법을 활용하여 각 건물의 에너지맵을 사용패턴 별로 그룹화 하여 개별 건물의 소비 특성을 파악할 수 있다. 또한 전체 캠퍼스의 에너지 맵핑은 피크 발생 예측순위를 확인할 수 있으며 각 시간의 건물별 사용량 에너지맵을 통해 관리가 용이하다. 본 연구에서는 에너지 절감가능성을 제시하기 위해 최적기동/정지 시간을 판단하는 알고리즘을 제시하였으며 해당 알고리즘을 적용할 경우 연세삼성학술정보관의 난방기간에 층별/날짜별로 많게는 5 시간에서 1 시간 까지의 개선을 보였으며 이는 전체 공조기 가동시간의 많게는 32.27% 적게는 5.6%에 해당한다.