Research on the Inﬂuence of Mobile Short Video Plat-form Recommendation Information Characteristics on User Stickiness

. The outbreak of the scientiﬁc and technological revolution has brought about the vigorous development of the digital economy. Many short video APPs have gradually come into people’s sight and become important tools for leisure and entertainment. At this time, improving user stickiness has become the key to their competitiveness and long-term development. As a major feature of short video platform, information recommendation plays an important role in the sticky behavior of users. Therefore, targeted research on the internal mechanism of platform recommendation information features for user stickiness will help to improve the platform user stickiness. Based on the stimulus organism response theory, this paper constructs a theoretical model of the impact of the characteristics of recommendation information on the user’s sticky behavior during the use of mobile short video, and investigates 336 mobile short video platform users through online questionnaires. The research shows that the accuracy, beneﬁt, interest and surprise of mobile short video recommendation information have a positive impact on users’ ﬂow experience and sticky behavior; The ﬂow experience has a positive impact on the sticky behavior of mobile short video users. This paper puts forward management suggestions for the management and operators of mobile short videos to help them optimize the design of their products, and recommends more accurate, proﬁtable, interesting and surprising information based on user characteristics to stimulate users’ ﬂow experience, thus improving users’ stickiness and helping enterprises achieve long-term development.


Introduction
Short video has become the main choice form of entertainment, consumption and shopping in the new era of mobile Internet due to its adaption to the life form of high information, people's highly fragmented entertainment time and the active pursuit of product information. However, "content is king" has always been the focus of the development of short video industry. The development of mobile communication technology makes the service content and functions of short video APP present the characteristics of diversification and richness.
As one of the key factors affecting the attitude, intention and behavior of APP users in the process of using the APP (Chang Yaping, Dong Xuebing, 2014) [1], the research on the information content characteristics of APP has become a hot topic in the domestic and foreign industry and academia in recent years. With the development of Internet communication technology, the information on the network presents a "blowout" growth trend, and people gradually transition from the blind area of lack of information to the contradiction of information overload. In order to help users obtain quality content efficiently and conveniently in the ocean of information, intelligent recommendation function emerges at the right moment and is applied in Zhihu, Weibo, Douyin, Taobao and other platforms. The recommendation information element in mobile short video is also one of its advantages. However, at present, most platforms including mobile short videos pay attention to the performance and action mechanism of the recommendation system while ignoring the quality of the recommendation information itself, resulting in the information recommendation on the platform is not completely satisfactory and laudable. Sometimes, the content launched has the problems of poor quality and information cocoon, which on the one hand will lead to the insufficient appeal to users' eyes. Users cannot be immersed in video content, resulting in poor experience. On the other hand, users will feel that the platform cannot meet their own value needs, entertainment needs, or knowledge acquisition needs, thus losing a large group of users, which is not conducive to the accumulation of traffic on the platform. Therefore, how to maximize the use of personalized information recommendation, so that it can attract users' interest and enhance the user retention rate of the platform, is the core issue that short video platforms need to face at present.
In this context, many scholars at home and abroad have studied the value of personalized recommendation. However, current studies mostly start from the technical perspective, hoping to improve the accuracy of personalized recommendation by optimizing the algorithm and model construction of personalized recommendation, so as to enhance the fit between product recommendation services and consumer preferences (Zhang Wu, 2022) [2]. However, there are few studies on the relationship between personalized recommendation and consumers, and most of them take purchase intention (Wang Jianming, 2021) [3] or adoption intention (Chen Meimei et al., 2020) [4] as the outcome variable. At the same time, in the process of research, it is more applied in the field of e-commerce and electronic library, a few are applied in wechat, mobile medical and other social media platforms, and the dabbing in mobile short video platform is very rare.
In view of the lack of existing research, this paper takes the mobile short video platform as the research background, focuses on the impact of the quality of recommendation information on user sticky behavior, and the entry point is more specific and micro compared with previous research, and innovatively introduces the heart flow theory in the research. Based on this, this paper studies various features of personalized system recommendation information, which provides guidance for enhancing user stickiness. From the perspective of platform operators and managers, it can help improve the pertinence of recommendation information from the product optimization design level, stimulate user flow experience to the maximum extent, improve user retention rate and user stickiness, and help the long-term development of mobile short video platform; Secondly, from the perspective of users, they can also get more interesting video content and better entertainment experience and value creation.

Features of recommendation information
Personalized recommendation information generated by the recommendation system is a process of displaying information, goods or services that users are interested in. However, there is no unified definition for personalized recommendation system. Research is generally conducted from the perspectives of technology and customers. At the technical level, it is expected to improve the accuracy of personalized recommendation by optimizing the algorithm and model construction of personalized recommendation, so as to enhance the fit between product recommendation services and consumer preferences (Zhu Yan, Lin Zannan, 2009) [5]. There are few studies on the perspective of customers. Personalized recommendation system is considered as a kind of service technology. The platform explores consumers' personal preferences and code of conduct according to various behavioral data of consumers, and then selects corresponding commodity or service information and sends it to consumers for selection (Pu Bin, 2016) [6]. It is designed to help customers quickly and accurately select the desired information from the complicated information screening. Based on the above views, the personalized recommendation system is defined in this study as an auxiliary decision-making system that provides short video information in line with customers' preferences and needs according to their personal characteristics, with the main purpose of recommending videos and service information in line with users' needs to different users. Meanwhile, the recommendation information studied in this paper is also based on the video information provided by the personalized recommendation system.
In summary, there have been many empirical studies on systematic recommendation information, but the research objects are mostly concentrated in the fields of e-commerce and digital library, and few involve the field of short video. In addition, numerous literature studies show that systematic recommendation information has an impact on consumers' psychology and behavior, which also provides support for the research of this paper. Therefore, this paper focuses on the information level of personalized recommendation system under the background of short videos, and refers to the research of Zhou Rongfu et al. (2016) [7], which divides the characteristics of recommendation information into four dimensions: accuracy, benefit, interest and surprise. Accuracy refers to whether the video information recommended by the algorithm recommendation system conforms to users' preferences; Benefit refers to the practical value of the video information pushed, which can help users in all aspects of life; Interest refers to the video information content is vivid, relaxed and humorous, and has enough new ideas and attractive places; Surprise refers to the content of the launched video information belongs to the category that they have never dabbles in, which is the mining of the potential needs of users.

Flow experience
The concept of flow experience was first proposed by Hungarian psychologist Mihaly Csikszentmihalyi (1975), who believes flow is the optimal experience that can bring happiness to people. Flow is defined as a state of involvement. When people are engaged in something they are interested in, they will forget the surrounding environment and lose self-awareness. No sense of time, into what is called the "artesian state" [8]. Nine aspects of challenge and skill, behavior and perception, goal, feedback, concentration, control, loss of self-consciousness, time distortion and self-purpose are given. In the field of customer behavior research, flow refers to the excited, happy and fulfilled psychological state experienced by customers in the process of consumption (Gong Xiaoxiao et al., 2019) [9]. As an important concept in positive psychology theory, flow emphasizes the study of human psychological phenomena and behaviors from a positive and constructive perspective, but excessive flow may also bring addiction or impulsive behavior.
Research shows that personalized recommendation service has a positive impact on users' willingness and behavior, but the intermediate factors are mostly perceived value, experience, trust, etc. Although a few subsequent studies have introduced the psychological factor of flow experience, there is a lack of detailed research on different internal factors of flow. The dimensional division after flow experience is basically based on Csikszentmihalyi's research.  combined the affective and cognitive value research in flow research and pointed out that flow experience mainly includes three dimensions, namely perceived control, perceived pleasure and mental concentration [10]. In addition, subsequent scholars used this method to divide dimensions and conduct related studies (Wang Liming et al., 2021) [11]. Therefore, this paper introduces the intermediary of flow experience, and uses this classification method to divide flow experience into perceptual control, concentration and enjoyment. Perceived control reflects the user's sense of self-control when using personalized recommendation services, including the control over the surrounding environment and activities. If the user has relatively high self-efficacy or is familiar with personalized recommendation services, the user will have a high perceived control. Concentration reflects the user's participation and immersion when using mobile services; Enjoyment reflects consumers' enjoyment and pleasure when using personalized recommendation services.

User stickiness
The research on user stickiness first appeared in the study of e-commerce. With the development of the Internet, user stickiness has been applied to a wider range of fields. Stickiness is generally studied from the perspective of websites and users. From the perspective of websites, stickiness is the ability of websites to maintain online customers and increase their browsing time each time (Judy, 2007). From the perspective of users, stickiness refers to the behavior and willingness of users to use specific websites repeatedly (Wu Xuanying, 2019) [12]. Researchers also found that user attitudes, trust and content quality of websites affect the formation of user stickiness [13]. Stickiness psychology and behavior are increasingly evident in Internet user groups. Many scholars have conducted researches on the stickiness of several Internet products such as new media, social networks, e-commerce platforms and mobile terminal applications, and defined stickiness from different dimensions, such as game stickiness, website stickiness and blog stickiness.
After summarizing the relevant research on user stickiness, it is found that when scholars select user stickiness as a variable, most of them focus on macro factors such as user psychological factors or system functions, while there is little research on micro factors of information itself, especially the research on information quality generated by recommendation systems. In addition, the intermediary factors are mostly experience, trust and perceived value, There is no important intermediate role of psychological factors in the whole process, such as a flow experience generated by the user's own stimulation. Based on the above studies, this study defines user viscosity from the perspective of users and combined with the current research background of mobile short video platforms as a kind of psychological dependence and repeated experience behavior formed by mobile short video users after they experience short video recommendation video information. Combined with the research background of mobile short video platform, starting from the stimulation of the recommendation information characteristics of the recommendation system, the introduction of flow experience focuses on the research of user's sticky behavior, so as to enrich and improve the relevant research of user stickiness.

Theoretical deduction and research hypothesis
The theory of "S-O-R" originates from environmental psychology, which regards the external environment as a stimulus that will affect people's psychological state and ultimately affect people's behavior (Fan J, 2020) [14]. Woodworth first proposed the SOR model in 1929, that is, the "stimulus organism response" model. Specifically, S in the model refers to the external stimulus that drives the response of the organism, R refers to the behavioral response generated by the organism, which is manifested as approaching or avoiding behavior, and O refers to the internal psychological state of the individual between the stimulus and the final behavioral response, including cognitive and emotional states. At the beginning, it was mainly to explain and analyze the relationship between environment and human behavior, and then more and more researchers applied it to investigate the impact of network factors on consumer response.
The main users of mobile short videos are users who watch short videos. They participate in the viewing process of mobile short videos on the basis of recommendation information. On this basis, this paper takes the features of mobile short video recommendation information as the stimulus of external variables, and the flow experience as an emotional and cognitive organism plays the role of intermediary variables. Users are stimulated by the different features of the recommendation information to stimulate the changes of psychological conditions such as flow experience, which ends with users' sticky reaction behavior.

Influence of recommendation information characteristics on user stickiness
The theory of use and satisfaction shows that consumers hold the control over the use of media, and consumers will conduct a comprehensive evaluation of media in the later stage of use. When consumers feel that the media can meet their own needs, they will be highly satisfied with the media, thus increasing their continued use. Cenftelli (2004) pointed out that in the research on the use of information systems, the promotion factors have positive attributes, indicating that the target system has high quality, which is conducive to attracting consumers to use and adhere to the system.
First of all, the platform will enhance users' loyalty and satisfaction by recommending short video content that meets their personal interests to users. Zhang (2015) pointed out that the adaptability of recommendation information to consumer goals and tasks would significantly and positively affect the long-term friendly relationship and use intention between consumers and offline businesses. Zhang (2018) reached a similar conclusion. When consumers feel that they can meet their own needs after using the media, they will be highly satisfied with the media, thus increasing the chances of repeated use. Secondly, from the perspective of psychology, people's pursuit of entertainment is spontaneous and innate. People tend to have interesting things, and boring information will soon make customers bored. The increase in the interest of information content can better fit the pressure relieving psychology of customers, improve the speed and effect of information dissemination and the affinity of enterprises, leave more profound impression on customers, and thus enhance the loyalty and stickiness to enterprises (Zhou Rongfu et al., 2016). This is also confirmed in the research on the background of WeChat official account (Li Xu, 2021) [15]. Thirdly, "profit seeking" makes users pay more attention to what they can get, and real benefits and usefulness can arouse users' interest, thus stimulating users' positive behavior and willingness in the next step. Xu Baojia (2019) [16] proposed in the research on user stickiness impact mechanism of UGC tourism mobile applications that content dimensions, including product usefulness, credibility and richness, have an important correlation between user stickiness and flow experience. Finally, generally speaking, user needs will change over time, and some users can't even detect their own interests and needs. At this time, it is necessary to use the user's behavior data to speculate the content that may be required and to launch surprising information, which can not only meet the current information needs of users, but also can mine the potential information needs and interests of users through modeling, and affect the consumer's recommendation adoption intention (Ricci, 2011) [17].
Based on the above discussion, relevant assumptions are proposed: H1: The feature of recommendation information has a positive impact on the user stickiness of mobile short video platform; H1a: The accuracy of recommended information features has a positive impact on the user stickiness of mobile short video platforms; H1b: The benefit of the recommended information features has a positive impact on the user stickiness of the mobile short video platform; H1c: The interesting features of recommendation information have a positive impact on the user stickiness of mobile short video platform; H1d: The surprise of recommended information features has a positive impact on the user stickiness of mobile short video platform.

The Influence of Recommended Information Features on Heart Flow Experience
Information content in the context of the Internet will affect the flow experience of consumers. Therefore, as a specific product of the Internet, the content quality of short video will also affect the flow experience of users (Liu Wenxia, Dong Yin, 2022) [18]. Meanwhile, the positive relationship between personalized recommendation information and consumers' emotional response has been supported by several studies. First, people are more likely to be persuaded and focused on information related to their own needs. When the short video platform has a high degree of recommendation fit, users will pay complete attention to the video content, feel as if their browsing behavior is under control and cannot be stopped, and get pleasure from the short video (Dong Chao, 2022) [19]. Second, the trigger mechanism for users to create an immersive reading experience is closely related to the value that the recommended content can bring. Li Fuda et al. (2019) believe that Douyin users will consider the practicability of the goods and services described when watching short videos [20]. In addition, the complexity of network information makes netizens have higher requirements for the usefulness of information display in the browsing process. Positive comments are often seen under the short videos with useful information display. Thirdly, in short video marketing, interestingness is an indispensable performance feature. Consumers in today's society should not only pursue practical value, but also enjoy the value of entertainment when shopping online, so as to achieve spiritual satisfaction and happiness [21]. Finally, surprise recommendation information can improve users' perceived information quality, reduce users' information search efforts, and increase their perceptual control. At the same time, when individuals are willing to browse surprise recommendation information, they are more likely to focus on it. At this time, surprise information will cause sensory stimulation, and customers' perceived pleasure will also be enhanced, thus generating flow experience (Congwen D, 2010) [22]. Mitas et al. (2018) believe that according to the goal alignment theory, tourists expect to obtain novel experiences and experiences in the process of tourism. When the novelty goal is achieved, positive emotional response will be triggered, and the sense of pleasure will be enhanced accordingly [23]. It can be seen that each dimension of the characteristics of recommendation information has different degrees of influence on flow experience, and the generation of flow experience will naturally bring about positive user behaviors.
Based on the above discussion, relevant assumptions are proposed: H2: Recommended information features have a positive impact on flow experience; H2a: The accuracy of recommended information features has a positive impact on perceptual control; H2b: The accuracy of recommended information features has a positive impact on mental concentration; H2c: The accuracy of recommended information features has a positive impact on pleasure; H3a: The benefit of recommendation information features has a positive impact on perceived control; H3b: The benefit of recommendation information features has a positive impact on mental concentration; H3c: The benefit of recommendation information features has a positive impact on pleasure; H4a: The interestingness of recommendation information features has a positive impact on perceptual control; H4b: The interestingness of recommendation information features has a positive impact on mental concentration; H4c: The interestingness of recommendation information features has a positive impact on pleasure; H5a: Surprise of recommendation information features has a positive impact on perceptual control; H5b: Surprise of recommendation information features has a positive impact on mental concentration; H5c: Surprise of recommendation information features has a positive impact on pleasure.

Effect of flow experience on user stickiness
In the context of the Internet, many studies have shown that flow experience will bring positive behaviors (such as stickiness or loyalty) (Ruth, 2001) [24]. Some scholars even found that flow experience has a stronger predictive power on behavior than satisfaction (Gao L, 2015) [25]. When an individual enters a state of flow experience, they enjoy it and want it again. Accordingly, relevant assumptions are put forward: H6: Heart flow experience has a positive impact on user stickiness; From the perspective of the internality of flow experience, scholars have conducted specific research on different dimensions. Zhang Hongxia (2021) conducted relevant empirical research on online games. The research results show that the immersion elements and immersion experience in the game experience process have a significant positive correlation with the user's willingness to use and flow experience, thus driving users to play repeatedly [26]. When users use mobile short videos, they can be attracted to the short videos and completely immersed in them. A high degree of concentration will make users satisfied with the use of the APP (Gao Xiaoqian et al., 2015) [27]. Liu Dan et al. (2018) pointed out that if learners feel happy and satisfied in the learning process, they will improve the continuous return frequency and stay time of the platform [28]. When consumers have flow, they can have a high sense of pleasure in the activities they are engaged in, and experience the integration of behavior and consciousness. This experience is mainly characterized by "presence" or "immersion", which may play a positive role in promoting consumers' impulsive buying behavior and sustainable behavior.
Based on the above discussion, relevant assumptions are proposed: H6a: The perceived control dimension of flow experience has a positive impact on user stickiness; H6b: The concentration dimension of flow experience has a positive impact on user stickiness; H6c: The pleasant dimension of flow experience has a positive impact on user stickiness; According to the above relationship research, the research model is constructed as shown in figure 1.
"presence" or "immersion", which may play a positive role in promoting consumers' In this paper, questionnaire survey is used to make an empirical study of the proposed model. The measurement items were based on the existing literature and adjusted appropriately according to the specific situation of the mobile short video context. The first part is the collection of personal information unrelated to variables, aiming to make the investigation of control variables simple and easy. The second part includes 29 questions, which is also the key part of the questionnaire. The questionnaire mainly measures the accuracy, benefit, interest and surprise of the recommended information in the research model, perceptual control, concentration, enjoyment, user stickiness and other variables. Except for the first part, all other variables were measured by Likert seven-point method, and the accuracy dimension was measured by Knijnenburg (2012) [29] and Song Hui (2011) [30]. The dimension of interest should refer to the scale of Wang Baibin (2013). [31]. The dimensions of fun refer to the scales of Lin et al. (2005) [32], Moon&Kim (2001) [33] and Barnett (1990) [34]. For the dimension of surprise, refer to Gan Zimei's (2020) [35] scale; for the measurement of flow experience, refer to   [36] scale; for the measurement of user stickiness, refer to   [37] and Paul's (1999) [38] scale. Then, combined with the characteristics of the mobile short video situation, the measurement items were modified and adjusted accordingly, and the measurement scale of this paper was finally formed. For specific measurement items, see the appendix.

Data Collection
This paper takes the mobile short video platform as the research context, aiming to explore the impact of recommendation information on users' sticky behavior in the platform, and mainly selects users of the mobile short video platform as the survey object. On this basis, the questionnaire in this paper is mainly distributed online and offline. The questionnaire star is used as a design tool to issue the questionnaire online, and it is spread through social networking tools such as WeChat, QQ, Zhihu, etc. Offline, the questionnaire is distributed in two large shopping malls with large traffic volume, Lifeng International Shopping Center and SEG International Shopping Center. 421 questionnaires are collected online, and 380 questionnaires are finally collected, There are 336 valid questionnaires, and the effective recovery rate is 88.4%. After collation, descriptive statistical analysis results are shown in the appendix. Specifically, in terms of gender, males accounted for 41.96% and females accounted for 58.04%, showing an equal ratio between males and females. In terms of age, people under the age of 18-39 accounted for 85.42% of the total sample. It can be seen that the user group of short video applications is relatively young. In terms of education status, among the total population participating in the survey, the population with bachelor's degree or above accounted for 66.96% of the total sample, while the population with college degree or below accounted for a relatively small number, indicating that the overall educational level of the population participating in the survey was high, which also increased the credibility of the survey results to a certain extent. Among the commonly used short video applications, 45.83% of people chose Tiktok, followed by 23.51% of Kwai and 16.96% of micro vision. Tiktok has become one of the most popular applications in the field of short video applications. For the time spent on short video applications, the average number of people who use short video applications for more than one hour every day accounts for 63.31%, which indicates that short video applications have become an important habit in people's daily life, and people rely heavily on short video applications.

Data analysis and result discussion
(1) Reliability and validity analysis of the scale CFA is used to assess internal consistency (reliability), project load (aggregation validity), and discriminant validity. Each project is modeled as a reflection of its assumed underlying structure. In the CFA model, these structures can change simultaneously, and the measurement model is evaluated using maximum likelihood estimates. According to the data in table 1, Cronbach's α coefficient is above 0.7, and the combination reliability (CR) is above 0.7, indicating good internal consistency of all test items. The AVE square root of each construct was greater than the correlation coefficient between the construct and other constructs, indicating good discriminative validity (see table 2).
For the judgment of discrimination validity, the square root of AVE is generally compared with the correlation coefficient between variables. It can be seen from table 2 that the AVE square root of all variables is greater than the correlation coefficient between this variable and other variables, which proves that there is good discrimination validity among the variables.
(2) Path analysis and hypothesis testing In this paper, the structural equation model is used to test the proposed model and assumptions. In the model, the independent variables include accuracy, profitability, interest and surprise. The intermediary variables include perception control, concentration and pleasure. The result variable is user stickiness. This paper uses AMOS18.0 for path analysis to obtain the relevant path coefficients and their significant performance. The results are shown in table 3. Table 3 shows that the accuracy, profitability, interest and surprise of mobile short video platform recommendation information have a significant positive impact on user stickiness (β =0.312, p<0.001; β =0.461, p<0.001; β =0.420, p<0.001; β =0.413, p<0.001), so the

The Mediation Effect Test of Heart Flow Experience
In this study, Baron&Kenny (1986) proposed the mediator effect test to study the mediation of flow experience. They pointed out that a variable (M) is an intermediate variable when it meets the following conditions: (1) there is a significant correlation between X and Y; (2) M will change with X; (3) Y will change with M; (4) If M is added, the previously significant correlation between X and Y is no longer significant. This shows that M has a full mediation. If the relationship between X and Y becomes weak but still significant. This shows that M has partial mediations. In this study, the independent variable X is the feature of recommendation information, the dependent variable Y is user stickiness, and the intermediary variable M is flow experience. The first three cases have been verified in the existing literature, and then the fourth case is verified. As shown in table 4 below, the regression coefficients between accuracy and user stickiness, interest and user stickiness, interest and user stickiness, surprise and user stickiness decreased significantly after adding the intermediary variable flow experience. To sum up, the flow experience has some intermediary effects.

Research conclusion
On the basis of existing research, this paper constructs and verifies the impact model of mobile short video recommendation information features on user stickiness, and draws the following research conclusions: (1) The characteristics of recommended information, including the accuracy, benefit, interest and surprise of information, have a significant positive effect on user stickiness. The higher impact on user stickiness is profitability and interest, and finally surprise and accuracy, which helps users to stay on this platform and browse videos repeatedly.
(2) The flow experience has a positive impact on its user stickiness. The flow experience reflects a kind of psychological state of people. When they are stimulated, they will enter a state of extreme enjoyment and generate flow. This state will make users stick to the platform and stay on the platform.
(3) The accuracy, benefit, interest and surprise of recommendation information characteristics have a positive impact on flow experience. When using the mobile short video APP, the video information recommended by the system with one of these characteristics will stimulate the user's flow experience, generate a strong sense of control and pleasure, and forget the time to immerse in the short video.

Management inspiration
The empirical research results of this paper have a certain enlightenment on the mobile short video platform to improve user stickiness based on the optimization of information recommendation services. The specific management enlightenment will be given mainly from two aspects: the emphasis on system recommendation information and the targeted optimization of recommendation information categories.
(1) Give enough attention to the role of the recommendation system The development of science and technology has brought about a surge in the amount of information, so that people are always faced with the search and filtering of information, as is the case in mobile short videos. Although users can obtain the video they want to watch through independent search, it will still bring a certain time cost, and it may be difficult to access the latest information, and fall into the dilemma of information cocoon room. At this time, the recommendation system has assumed the responsibility of solving the dilemma. By accurately recommending relevant videos for users, it has brought convenience to the platform, bloggers and users. Therefore, the operators and managers of the short video platform should attach importance to the information recommendation system to help the platform master the traffic password and stand out from many competitors.
(2) Timely optimize the information recommendation system From the perspective of platform operators and managers, we should, according to the research conclusions of this paper, focus on the four characteristics of accuracy, benefit, interest and surprise in the information recommendation process, and optimize the design of our own information recommendation system.
First of all, the platform should pay attention to the benefits of the information in the recommendation information, and the platform should focus on strengthening the benefits of the content when pushing information. Only when customers can fully feel the existence of information benefits and meet their needs, can they stimulate the stickiness of users. Secondly, interesting video content and video atmosphere will bring a pleasant mood to users, so as to better fit the pressure relief psychology of users, enable users to have a positive psychological effect on the platform, and increase users' favor and stickiness to the platform. Thirdly, it is necessary to deeply explore the potential interests and needs of users to avoid the situation of "information cocoon room". Some users cannot even detect their own interests and needs. At this time, it is necessary to use the user's behavior data to speculate the content of possible needs. Finally, we should dynamically combine users' interests and needs to recommend relevant content, which will be more in line with users' appetite, but we should balance accurate and diversified recommendations, expand the scope of recommended content, and avoid causing users' aversion and aesthetic fatigue.

Future outlook
Specifically, this paper can optimize from the following aspects: First, increase the sample size and research scope. The research object of this paper is gathered in the area centered on Xi'an University of Technology. In addition, it is friends of the author's social circle, and the research scope is relatively small. The following research should expand the scope of investigation and pay attention to the wide distribution of all age groups and knowledge levels to improve the scientificity and reliability of the research.
Second, the adjustment variable is added to the existing model to deepen the research and make the research path clearer. For example, in future research, privacy, platform involvement, etc. can be considered as moderating variables in the research to provide more accurate suggestions for enterprises.