(A) Study on the factors influencing the cooperation intention between Chinese enterprises and Korean enterprises

Zheng MeiLin 호서대학교 대학원 2021 국내박사

This paper aims to explore the factors that influence the cooperation intentions of Chinese and Korean enterprises. First by studying the existing literature, it can be concluded that technological innovation capability, network capability, national image, organizational trust, and government support will impact the intentions to cooperate. Moreover, organizational trust and government support play an intermediary and moderating role which influences the intentions of Chinese and Korean enterprises to cooperate. A questionnaire was conducted among the representatives of 323 Chinese enterprises by combining online and offline survey methods. The results of empirical analysis show that technological innovation capability has no significant impact on cooperation intention. However, network capacity, national image, and organizational trust have a significant and positive impact on cooperation intention. Thus, non-technical factors are more important determinants than technical factors. Besides, after analyzing the mediating effect of organization trust, it is found that the three variables of technological innovation capability, network capacity, and national image all have a significant and positive influence on the mediating variable trust. In addition, the moderating effect of government support shows that only the interaction with the national image can be statistically significant. Moreover, the empirical results show that Chinese enterprises with five to ten years of cooperation experience with Korean enterprises are more willing to cooperate with Korean enterprises than other enterprises. According to the results of this study, we believe that in order to promote cooperation between Korean and Chinese enterprises, several efforts should be made to improve the overall national image, strengthen public relations, and strive to increase mutual trust. The limitations of this research are as follows. First, the objects of this study are distributed in some specific fields, and they are not representatives of all Chinese enterprises. Second, the proportion of enterprises from the power and electronics industries is too high, which may lead to a deviation of the overall research. Third, the respondents are selected as research objects when they attend specific exhibitions, but not the actual representatives of each enterprise, which is a significant limitation.

A study on the impact of live streaming marketing of internet celebrity on the purchase intention

샤오이 호서대학교 대학원 2022 국내박사

Internet celebrity economy and webcasting, as well as the research status of consumers' purchase intention, combined with the empirical research of some research scholars, and the Internet celebrity live streaming shopping. Conducted in-depth interviews with consumers, put forward the research hypotheses and models of this paper, designed questionnaires and scales, and analyzed the influencing factors of internet celebrity live streaming on consumers' purchase intentions. Questionnaires were distributed through Questionnaire Star and on- site, and 287 valid questionnaires were recovered through SPSS for descriptive statistics, reliability analysis, validity analysis, correlation analysis, and regression analysis. The following conclusions were obtained: (1) Among the influencing factors of internet celebrity live streaming on consumers' purchase intention, the preferential situation and interaction of products directly and positively affect consumers' purchase intention, and the preferential situation of products has a significant impact on purchase intention. The direct effect is greater than the direct effect of interaction on purchase intention. (2) Trust is a very important mediating variable in the influence of various dimensions of internet celebrity live streaming on consumers' purchase intention. (3) Perceived value is another important mediating variable between internet celebrity live streaming and consumers’ purchase intention. At the end of the paper, combined with the research process and research results, some management suggestions are put forward for Internet celebrity anchors, live streaming platforms, e-commerce companies and consumers, and summarizes the research shortcomings and prospects of this paper. With the rapid development of the Internet, the popularization of intelligent terminals and people's full use of fragmented time, the new marketing method of internet celebrity live streaming has developed rapidly. This industry has been transformed into an emerging industry by virtue of the application of technology, with a large amount of capital investment and good operation. There are huge challenges hidden behind the Internet celebrity live streaming industry. In today's era, the research on the purchasing decisions of consumer groups under the Internet celebrity live streaming is particularly important.

The Impact of E-Commerce Live Streaming Interaction on Consumers' Purchase Intentions : TikTok live streaming platform as an example

류스하이 호서대학교 대학원 2022 국내박사

With the booming development of China's Internet technology and the increasing popularity of intelligent terminals, as a new retail approach to live e- commerce began to be attached to and become a must for businesses and platforms, the consumption of live shopping began to be favored by the majority of consumers, because compared with the traditional online shopping model, the interaction between live appears to be more diversified, shopping options are also more diverse. Even so, the interaction during the live shopping process and the customer's purchase intention is often unsatisfactory, and there are always a variety of problems. With the growing popularity of live e-commerce, how to improve customer satisfaction in live e-commerce to retain new and old customers, has become a real problem faced by live e-commerce platform and platform merchants. In the environment where e-commerce live streaming is in full swing, research on the interactivity, consumer trust and consumer purchase intention in e-commerce live streaming has not received enough attention in the academic world, and there is a lack of more thorough research. Based on previous research on traditional online shopping scenarios, this paper reviews and summarizes the literature and research related to interactivity, consumer trust, and consumer purchase intention. This paper focuses on the TikTok live streaming platform, and refines the interactivity in TikTok live streaming into three elements of interesting, responsiveness, and proximity, and subdivides consumer trust into two dimensions of cognitive trust and emotional trust. On this basis, a model of consumers' purchase intention was constructed and research hypotheses were proposed. Subsequently, the research data were collected using questionnaires, and the following empirical findings were obtained using descriptive statistical analysis, validity and reliability tests, discriminant validity analysis, reliability analysis, structural equation model construction and structural equation fit tests to analyze the relationship between interactivity, consumer trust and consumer purchase intention. 1. There is a significant positive effect of interesting,responsiveness and proximity on cognitive trust; There is a significant positive effect of interesting, responsiveness and proximity on Emotional trust 2. Cognitive trust and emotional trust have significant positive effects on customer purchase intention. Finally, based on the findings of the study, this paper puts forward targeted suggestions on how to improve consumers' purchase intention from the perspective of interactivity, in order to effectively improve customer experience and increase consumers' purchase intention in TikTok Live.

Role of heat shock transcription factor 1 in periodontal tissue in vivo and cells In vitro

김현수 경희대학교 대학원 2017 국내박사

Heat shock transcription factor 1 (HSF1)은 세포의 열충격반응 (HSR: heat-shock response)을 조절하는 전사인자로서 세포를 스트레스로부터 보호하거나 세포대사 및 기능에 관여하는 유전자에도 영향을 미치는 것으로도 알려져 있으나 치주조직 및 골대사에서 역할은 밝혀져 있지 않았다. 본 연구의 목적은 치주조직 및 치주인대세포에서 HSF1의 역할을 규명하기 위해 백서 발생치주조직에서 발현을 알아보고, 치주인대세포의 조골 및 파골 분화에 미치는 영향과 관련된 신호전달 기전을 밝히고자 하였다. 출생 후 1주, 2주와 4주 백서조직에서 면역조직화학적 방법으로 HSF1의 발현을 관찰하였고, 인간치주인대 세포에서 분화는 alkaline phosphatase (ALP) activity, RT-PCR, alizarin red staining과 Western blotting을 시행하여 평가하였다. 생쥐 골수기원대식세포에서 receptor activator of NF-κB ligand (RANKL) 유도 파골세포로의 분화를 tartrate-resistant acid phosphatase (TRAP) 염색을 통해 확인하였다. 치근 형성 전단계 (postnatal day 7, PN7)의 백서 치주조직에서 치낭세포, 법랑모세포와 골모세포에서 HSF1 발현이 관찰되었다. 치근 형성 단계 (PN14)와 치아 맹출 단계 (PN28)의 백서 치주조직에서 골모세포, 백악모세포와 치주인대세포에서 HSF1 발현이 관찰되었고 치주조직 발생단계에 따른 HSF1 발현 양의 차이는 없었다. 니코틴과 lipopolysaccharide (LPS)로 자극된 치주인대세포에서 HSF1의 발현의 증가가 관찰되었다. 니코틴과 LPS에 의해 억제된 ALP activity, osteoblast marker gene인 Runx2, OPN, OCN의 억제된 발현은 HSF1 억제자인 KRIBB11와 HSF1 siRNA에 의해 회복 되었다. 니코틴과 LPS에 의해 상향조절된 osteoclastic gene중 tumor necrosis factor alpha (TNF-α), Interleukin beta (IL- ß), RANKL과 macrophage-colony stimulating factor (M-CSF)은 KRIBB11와 HSF1 siRNA에 의해 억제되었다. KRIBB11와 HSF1 siRNA는 파골세포 전사인자인 c-Fos와 NFATc1를 차단하고, 파골세포의 분화를 감소시켰다. 본 연구는 처음으로 백서 발생치주조직에서 HSF1이 치낭세포, 법랑모세포, 골모세포, 백악모세포와 치주인대세포에서 발현함을 밝혔으며, KRIBB11와 HSF1 siRNA가 니코틴과 LPS로 자극된 치주인대세포 모델에서 Smad1/5/8, Toll-like receptor (TLR), Akt, mitogen-activated protein kinases (MAPKs), nuclear factor-κB (NF-κB) 경로를 통해 파골세포 분화의 억제와 억제된 조골세포 분화의 회복의 역할을 한다는 것을 밝혔다. 이에 따라 HSF1은 치주조직의 재생과 교정적 치아 이동과 관련된 골대사의 표적 분자로 이용될 가능성이 있을 것으로 사료된다.

합성곱 신경망을 이용한 딥러닝 기반의 토지피복 분류와 변화탐지

조원호 서울대학교 대학원 2022 국내석사

The land cover map plays an important role in land management and planning. The land cover map of Korea is produced and distributed by the Ministry of Environment. It is used as an auxiliary data in various research as well as to grasp the current status of earth. However, the land cover map produced by Ministry of Environment has problems of classification accuracy and update cycle. Land cover map which has low classification accuracy can distort the results of research. The limitation of the land cover map is because of its classification method. This study propose a deep learning based method as a land cover classification. The existing method used a parametric models. However, there were many strict assumptions such as independent assumption and normal distribution assumption. Non-parametric model presented in machine learning field can overthrow these unrealistic statistical assumptions. This study intends to expand the land cover classification method, which has been focused on parametric models, to a deep learning-based method. And furthermore this research present a deep learning-based land cover change detection method. To prove the effectiveness of deep-leaning based method, this study present a convolutional neural network model for land cover classification. It includes the design work of the architecture and experiment of classification. In addition, this study examines the performance of the model according to the various input data. Case of land cover change detection, this study proposes a fully convolutional neural network model. The designed change detection model is tested in two cases. Key results of the study are as follows. First, the convolutional neural network was grafted onto land cover classification and its effectiveness was verified through experiments. The land cover classification model showed high classification accuracy. The importance of the training data was emphasized by presenting the difference in results according to the input data. In particular, the possibility of transfer learning using pre-trained parameters was shown through experiments, and it was confirmed that parameters trained from images which have noting to do with land cover data shorten the learning time and improve classification performance. Second, the model based on the fully convolutional neural network effectively performed change detection in the case area. Through this model, the detection and segmentation of land cover can be conducted simultaneously and its result can be expressed in continuous from. This study is meaningful in that it has demonstrated the effectiveness of land cover classification and change detection using convolutional neural network, which had not received much attention in South Korea. It is also meaningful in that it has been confirmed it can be an alternative to overcome the limitations of the existing method and put it into practice. Given such suggestions, it is expected that the results of this study can be used as a new method in the field of remote sensing with the theme of land cover. 위성영상 자료를 토대로 제작되는 토지피복도는 국토 관리와 계획에 중요한 역할을 하며 다양한 연구에 활용되는 자료이다. 대한민국의 토지피복도는 환경부에서 제작해 배포하고 있다. 그러나 환경부의 토지피복도는 정확도가 낮고 갱신주기가 길어 자료를 활용하는데 한계가 있다. 낮은 정확도는 이를 활용하는 연구의 결과를 왜곡시킬 수 있으며, 반복적인 자료 수집이 어려워 토지피복도를 활용한 시공간 분석의 효용성을 떨어뜨린다. 환경부의 토지피복도가 가지는 한계는 토지피복 분류의 방법에서 발생하는 문제점에 기인하며, 이에 토지피복 분류 방법에 대한 새로운 방법론이 요구되고 있는 실정이다. 본 연구에서는 토지피복 분류방법으로 딥러닝(deep learning) 기반의 방법을 제안한다. 기존의 토지피복 분류 방법은 매개변수 모형(parametric model)을 사용한 방법들로 설명변수의 독립성, 자료의 정규성 등 엄격한 통계적 가정이 많았다. 기계학습(machine learning) 분야에서는 이러한 비현실적 가정과 사전에 분포 형태를 강제하지 않는 비모수 모형(non-parametric model)들이 다양하게 제시되어왔다. 본 연구는 그간 매개변수 모형에 집중되었던 토지피복 분류 방법을 딥러닝 기반의 비모수 분류 방법으로 확대하고 나아가 딥러닝 기반의 토지피복 변화탐지 방법을 제시하고자 한다. 딥러닝 모형 중 합성곱 신경망(convolutional neural network) 기반의 토지피복 분류 방법을 제시하며, 그 효용성을 실증하기 위해 토지피복 분류를 위한 합성곱 신경망 모형을 설계하고 분류 실험을 하였다. 아울러 합성곱 신경망 모형을 설계하여 모형의 우수함을 판명하는 것에 더해, 모형이 투입 데이터에 따라 어떠한 결과차이를 가지는지 세 차례의 실험을 통해 파악했다. 토지피복 변화탐지의 경우 완전합성곱 신경망(fully convolutional network) 모형을 설계하여, 토지피복의 인공 구조물 변화 그리고 토지피복의 자연환경적 변화의 사례에 대한 변화탐지를 실험했다. 연구의 결과는 다음과 같이 요약된다. 첫째, 합성곱 신경망을 토지피복 분류에 접목시켰고 실험을 통해 그 효용성을 검증하였다. 본 연구에서 설계한 토지피복 분류 모형은 높은 분류 정확도를 보였다. 토지피복 분류에 있어 모형의 우수함뿐만 아니라 투입하는 학습데이터에 따른 결과차이를 함께 제시해 학습데이터의 중요성을 역설하였다. 특히 사전학습 파라미터(pre-trained parameter)를 사용한 전이학습(transfer learning)의 가능성을 실험을 통해 보였으며 토지피복과는 무관한 이미지로 학습된 파라미터가 학습시간을 단축시키고 분류성능을 높인다는 것을 확인했다. 둘째, 연구에서 설계한 완전합성곱 신경망 기반의 모형은 사례지역에 대하여 변화탐지를 효과적으로 수행했다. 완전합성곱 신경망의 적용을 통해 변화의 탐지와 분할을 동시화해 효율성을 제고하였으며, 보다 연속적인 변화탐지 결과물을 보여주었다. 본 연구는 국내에서 주목받지 못했던 합성곱 신경망을 이용한 토지피복 분류와 변화탐지의 효용성을 실증하였으며, 기존 방법의 한계를 타개하고 실용하기 위한 대안이 될 수 있음을 확인하였다는 점에서 의미가 있다. 이에 본 연구의 결과는 향후 토지피복을 주제로 한 원격탐사 분야에서 새로운 방법으로 활용될 수 있을 것으로 기대된다.

머신러닝 기반의 동작인식을 수행하는 저전력 센서 허브를 위한 데이터 압축 및 전력 관리 기법

이승진 세종대학교 대학원 2018 국내석사

최근 다양한 무선통신기술의 발전과 디바이스의 소형화 기술을 기반으로 하여 IoT(Internet of Things)라 불리는 사물인터넷 기술이 보편화되고 있다. 사물인터넷 기술은 현존하는 IT(Information Technology) 기술의 집약체로 다양한 분야에서 사용되고 있으며, 특히 주위 환경이나 디바이스의 상태를 센서를 통하여 감지하고, 무선으로 전송하는 형태의 디바이스가 상당수를 차지하고 있다. 무선 환경에서 작동하는 이러한 기기들은, 주요 에너지원으로 배터리를 사용한다. 그러나 디바이스에 장착된 배터리의 용량은 한정되어 있기 때문에 전력효율을 극대화하는 저전력 기술이 상당히 중요하다. 센서 허브 구조는 센서 데이터를 높은 전력 효율로 처리하기 위하여 제안되었다. 센서 허브라는 센서 처리에 특화된 연산유닛을 추가로 사용함으로써, 메인 프로세서가 슬립 상태에서 깨어나는 빈도를 줄이게 되어 높은 전력 효율을 달성한다. 본 논문은 센서 허브 구조 내부에서 전력절감을 실현할 수 있는 방법으로, 수집되는 센서 데이터의 작은 차이를 압축하여 처리함으로써 센서 허브의 동작시간을 줄이고 메모리 효율을 증대시킬 수 있는 압축 기법을 제안한다. 또한 압축 기법을 기반으로 디바이스에 연결되어 있는 센서들의 전원 상태를 동적으로 변경하는 전력 관리 기법을 제안한다. 제안하는 기법들을 검증하기 위하여 SVM(Support Vector Machine)기반 동작 인식 어플리케이션을 사용했으며, 시뮬레이션을 통하여 검증하였다. 그 결과, 어플리케이션 수행 결과의 정확성 변화가 없으면서도, 센서 허브의 깨어나는 빈도와 데이터 전송량이 78% 감소하였으며, IMU(Inertial Measurement Unit)이 소비하는 전력의 약 59%를 절감할 수 있다. Recently, low-power IoT (Internet of Things) devices are equipped with a sensor hub which is a dedicated processor for processing sensor data to save power required for sensor data collection and processing. Furthermore, research is being performed on the low-power architecture of the sensor hub to reduce the power consumption of the sensor hub itself as well as to utilize the sensor hub in low-power systems. This thesis proposes a sensor data compression mechanism based on the amount of change of sensor input data and a power management scheme for various sensors used for the motion recognition application. The effectiveness of the proposed compression mechanism and sensor power management mechanism are verified based on the experimental results. The experimental results confirmed that the proposed compression mechanism and power management scheme reduced the wakeup count of the sensor hub core and the amount of data transmitted to the core by about 78% compared to the conventional data buffering structure respectively, and the power consumption of the IMU (inertial measurement unit) is reduced by about 59%.

Exploring the impact of contradictory online reviews with different structures on consumer perceptions from the perspective of attribution theory - a moderated mediation model

황방한 호서대학교 일반대학원 2022 국내석사

In the era when online shopping is gradually becoming the mainstream shopping method, online reviews are a reflection of consumers' subjective consumption feelings after consuming products or experiencing services, so online reviews have become an important standard for consumers to measure the online reputation of products. With the emergence of the additional review function, consumers can get additional review opportunities within a certain time frame after making initial reviews, thus reflecting the post-purchase experience more truly. Since such reviews are a delayed experience after the consumer has used the product or service, they imply that the purchaser has a more comprehensive and in-depth understanding of the product and the service provided by the merchant. In addition, initial and additional reviews by the same reviewer in a contrasting relationship are more likely to reveal similarities and differences in opinions than a single review, and have greater reference value for consumers. Online reviews containing additional reviews can be classified into consistent online reviews and contradictory online reviews based on the same and different directions from the initial review. Contradictory online reviews, as a type of online reviews, are an important channel for consumers to obtain product information in the online shopping environment. There is a large amount of literature and studies that illustrate that contradictory additional reviews have a stronger impact and are more valuable to study than consistent additional reviews. Most of the current research on contradictory online reviews focuses on the comparison of contradictory online reviews with consistent online reviews, but there is a lack of discussion on the different structures of contradictory online reviews in real life. What are the different effects on consumers' perceived risk and perceived usefulness of reviews when initial and additional reviews appear in different orders and produce contradictions? Few scholars have investigated this aspect. Current studies lack research on the structure of Contradictory online reviews and the perceived mental activity of observers, and there is a lack of insight into the case where the observers' own mental changes and the inherent properties of the goods are used as variables. Therefore, based on the theoretical foundations such as the Elaboration Likelihood Model (ELM) and the Information Adoption Model (IAM), this study takes contradictory online reviews of different structures and consumer perceptions as the main research objects, takes attribution to reviews at the consumer psychological level as the entry perspective, and uses product involvement as the moderating variable of this process, so as to explore the effect of contradictory online reviews of different structures on consumer This paper explores the impact of different structures of contradictory online reviews on consumers' perceived risk and perceived usefulness. This paper explores the following three questions: 1. Does the influence of contradictory online reviews on consumers' perceptions differ across different structures; 2. Does attribution to reviews play a mediating role in the influence of contradictory online reviews on consumers' perceptions across different structures; 3. Does product involvement play a moderating role in the different structures of contradictory online reviews influencing consumer perceptions through attribution to reviews?

Research on the Trust and Conformity with Buying Internet Celebrity Goods : Perspective of Planned Behavior Theory

Dai Chuangang 호서대학교 대학원 2021 국내박사

Online shopping has become the main way of Chinese consumers' daily consumption. In the e-commerce economy, the rapid development of online celebrities and the spring tide of online shopping have given birth to the popular of online celebrity economy. Online celebrity economy is a new economic phenomenon, and there are few related research literatures. The main purpose of this paper is that through the analysis of online celebrity economy, it is expected to find out consumer purchase traits when they consume online celebrity goods. If the conformity and trust trait exist, what kind of impact does they have on consumers' purchase intention. Based on the traditional planned behavior theory, this paper expands the theory, including the psychological characteristics of consumers' conformity and consumers' trust in the shopping platform into the model. Together with the attitude of consumers, the subjective norms of consumers and the consumers' perceived behavior control, they are regarded as the independent variables in the model, and the consumers' intention to purchase online celebrity goods is taken as the dependent variables. The significance, intensity, and direction of the influence of independent variables on dependent variables are studied. At the same time, gender variables as demographic variables is also included in the model, and the moderation effect of gender on the influence between independent variables and dependent variables is analyzed. Based on an effective questionnaire of 369 people, we obtained the first-hand survey data, and conducted basic statistical analysis and structural equation model analysis on the data. Data show that both consumer conformity and trust have a positive impact on consumers' purchase intention. It provides a reference for e-commerce enterprises to strengthen daily management. At the same time, it also provides consumption guidance for consumers who buy online celebrity goods, so that they can make more rational choice and purchase when facing online celebrity goods. The theoretical significance of this paper lies in the further expansion of the theory of planned behavior, which further demonstrated the openness and wide applicability of the theory of planned behavior. It is proved that the consumer conformity psychology has a positive effect on the purchase of online celebrity goods, and this influence is formed by the complete indirect effect of consumer attitude. In the same way, it also proves that consumer trust has a positive impact on online celebrity goods, and this effect is formed through the complete indirect effect of consumer attitude. At the same time, we also find that gender, which have strong moderation effect in other literatures, have no evidence of moderation effect in this model context, besides gender has significant moderation effect on the relationship between attitude and purchase intention.

(A) Moderated Mediation Study of Short Video Newsfeed Ads, Perceived Value, and Purchase Intention Based on SOR Theory

런가우페이 호서대학교 대학원 2021 국내박사

With the iterative upgrade of mobile Internet technology, mobile short video has become the mainstream means of information dissemination. As an important carrier of the commercial value of mobile short video, short video newsfeed ads are reshaping the process of content production and online marketing. What are the influencing factors of short video newsfeed ads on purchase intention? What is the effect mechanism? To explore the influencing factors and mechanism of the conversion of short video newsfeed ads into purchase intention, this study focuses on mobile short video in China, based on the Stimulus-Organism-Response (SOR) theory, from the three dimensions of newsfeed ads, platforms and users of the short video, constructed a moderated mediation model using perceived value and advertising rewards as mediating variable and moderating variable respectively, and tested the moderated mediation model using the data collected by the questionnaire. The empirical results show that perceived value (PV) has significant partial mediating effects in the positive effect of situated advertising (SA), celebrity endorsement (CE), personalized recommendation (PR), user interaction (UI) on purchase intention (PUI), and advertising rewards (AR) significantly positively moderated these mediating effects; platform satisfaction (PS) significantly positively affects purchase intention, but user perceived intrusiveness (PEI) has a significant negative effect on purchase intention. At the same time, the analysis of variance of short video usage behavior show that short video users preferred the advertising type of combination of professional generated content and user generated content, and users who preferred this type of advertising had significantly higher purchase intention; users who used three short video apps, namely Tik Tok, Volcano and Kuaishou, and users who spent more than 4 hours each time have significantly higher purchase intention for newsfeed ads products; users who have commented on short video newsfeed ads have significantly higher purchase intention than those who have not commented on them; short video users who have clicked the purchase links and given thumbs up have significantly higher purchase intention than those who do not. These research conclusions provide a theoretical basis and practical implications for advertising operators to design short video newsfeed ads, for advertisers to carry out advertising marketing and for short video operators to optimize platform design.

An Empirical Research on the Influencing Factors Model of Mobile Social Media Fatigue and Negative Use Behavior in China

Wang Binbin 호서대학교 대학원 2020 국내박사

[Purpose / Significance] Theoretical model analysis was conducted on the influence of users' own factors on behavioral intention of mobile social media fatigue; it is helpful for enterprises or related organizations to improve the business and public services of social media tools from the perspective of users themselves. [Method / Process] Based on the research paradigm of "individual perception-use attitude-behavior willingness-behavior", Combining theory of planned behavior, privacy calculation theory, flow theory and process incentive theory, the influence factor model of mobile social media fatigue behavior was constructed and the applicability of the model was tested by empirical research methods. [Result / Conclusion] Data analysis results show that, self-efficacy, user habits and flow experience have a significant positive effect on mobile social media use intensity. Privacy concerns have a significant negative effect on the mobile social media use intensity, but the effect is weak; Mobile social media use intensity has a significant positive effect on mobile social media fatigue; Mobile social media fatigue has a significant positive effect on the negative use behavior of social media, and different negative use behaviors have different influence intensity. The variable of process incentives has a negative moderating effect on the relationship between mobile social media use intensity and mobile social media fatigue.