ABSTRACT. 在科技引领时代发展的今天，人工智能已逐渐成为新一代科技革命的核心关键技术，是数字经济时代产业变革的重要驱动力量，深刻影响全球人类的生产生活，以及各国的政治、经济、文化发展。本报告主要从人工智能的兴起和发展、人工智能第三次浪潮的特点、人工智能技术前沿及应用、人工智能最新研究进展等方面，介绍数字经济时代背景下人工智能技术的最新研究及进展情况，对人工智能技术和相关产业发展进行讨论。

ABSTRACT. 围绕如何应用海洋渔业大数据技术提升传统海洋渔业，实现渔业现代化，助推海洋渔业产业健康快速发展，以多模态海洋渔业大数据综合服务平台构建为目标，以海洋渔业大数据感知、获取、分析等技术为重点，讨论海洋渔业大数据平台架构、海洋渔业环境监测数据分析与应用、海洋渔业文本数据的分析与应用、海洋渔业图像及视频数据的分析与应用等内容，最后总结多模态体系化海洋渔业大数据综合服务平台研究面临的挑战。

ABSTRACT. 人工智能技术与众多行业应用的融合如火如荼。当前，在线教育，作为人工智能应用示范的重要场景之一，正在向智慧教育演进。受此需求牵引，面向在线教育的智能导学成为一个重要的研究方向，具体针对个体自主在线学习过程中存在的“认知过载易迷航”问题，开展学习路径规划、学习资源个性化推荐、智能助教辅导答疑等方面的关键技术研究。本次报告主要介绍智能助教自动问答的研究背景和基本概念，并汇报团队在不可回答问题归因和时序约束知识问答等方面的工作进展。

ABSTRACT. 如何利用知识增强来设计一个兼顾用户体验和系统推荐效率的推荐系统呢？在报告中我们将讨论如何利用多行为来更好地完成用户和项目的嵌入，以更加全面的反映用户的独特偏好和项目的特征信息。传统的推荐技术通常侧重于对用户和项目之间的单一类型行为进行建模。然而，在许多实际的推荐场景（例如，社交媒体、电子商务）中，在用户-项目关系中存在多种类型的交互行为，例如点击、标记为收藏以及添加到购物车等。所以，如何充分利用多行为信息进行推荐对现有系统非常重要。现有的多行为推荐模型主要面临如下两个挑战：（1）如何处理用户个性化行为多样性，利用多行为依赖关系来捕获个体偏好；（2） 如何处理目标行为监督信号稀疏导致的推荐不足的问题。 针对以上问题，我们提出了一个知识增强多行为对比学习框架（KMCLR），包括两个对比学习任务和三个功能模块，以分别应对上述挑战。特别地，我们设计了多行为学习模块来提取用户的个性化行为信息以用于用户嵌入增强，并利用知识增强模块中的知识图来推导出项目的鲁棒知识感知表示。此外，在优化阶段，我们还对用户多行为之间的粗粒度共性和细粒度差异进行建模，以进一步提高推荐效果。在三个真实世界数据集上的大量实验和消融测试表明，提出的KMCLR方法优于各种最先进的推荐方法，验证了我们方法的有效性和关键组件的合理性。

ABSTRACT. 区块链食品追溯系统可以改善食品供应链中信息不对称的问题进而促进食品安全状况的提升，对消费者系统使用意愿及其影响因素的研究有助于促进系统的推广应用。本文在UTAUT模型的基础上结合区块链食品追溯系统的特点构建了系统使用意愿模型，运用结构方程模型进行了实证分析。研究结果表明：感知透明度、社会影响和便利条件均正向显著影响使用意愿；感知透明度对男性消费者使用意愿的影响更大，而便利条件对女性消费者使用意愿的影响更大；不安全食品购买经历和食品安全意识都显著正向调节感知透明度与使用意愿的关系。

ABSTRACT. 用户满意度是反应App质量的重要标准，版本更新是开发者满足用户需求的重要手段，但是受到用户需求多样性与开发者资源有限性之间矛盾的制约，开发者不能实现用户的全部需求，因此确定关键用户需求及其优先级是有意义的。本文从用户需求的视角出发，应用Kano模型确定用户关键需求，探究版本更新对用户需求的满足程度对用户满意度的影响，从而识别需求的优先级，为开发者提供更精确的意见参考。

ABSTRACT. 本文通过分析新冠肺炎爆发前后的在线评论数据集及患者向医生咨询的线上会诊记录，采用将词共现分析与LDA模型结合的CO-LDA模型进行评论主题挖掘，使用决策树算法及 libsvm 对各主题维度的子句进行情感分析和情感倾向值计算，探索疫情爆发前后患者就医行为的改变以及疫情的爆发如何对患者的情感倾向和就医体验产生影响。主要研究发现是：疫情爆发后，在线医疗平台每日在线咨询量显著增加，其中咨询量最多的科室为内科，来自湖北省的咨询量显著增加。疫情爆发后，患者对线上问诊就医体验满意度提高，对线下问诊就医满意度降低。研究结果能够帮助在线医疗平台及医院了解患者关注主题及情感倾向，针对性地对就医服务流程进行改进，改善疫情期间患者的就医体验。

ABSTRACT. Due to the weak effect of traditional advertising, more and more brands use product placement forms such as variety show title or product sponsorship. In the past, the research on brand implantation mainly focused on the implantation of physical products in movies, games and traditional TV programs, as well as the morality and disclosure of physical products, using the methods of questionnaires and surveys, and paid less attention to the difference in the impact of physical and virtual product forms on brand memory in the implantation of new media - online variety show brands, as well as the memory effect of matching and improving the implantation location. In this paper, using eye tracker and skin electricity multi-modal physiological data, field experiments and questionnaire survey methods, we objectively explore the influence and adjustment of the shape and location of the brand implanted products on brand memory in the real online variety show environment. The research found that: in the same program, compared with virtual products, consumers have significant awareness of physical product brand implantation, which makes brand memory stronger. Marketing personnel can adopt scenario design strategies to reduce the difficulty of information processing, so as to improve the effectiveness of virtual product brand implantation.

ABSTRACT. 本文依据压力源-负担-结果框架，对MOBA类网络游戏用户中辍行为的影响因素及其作用机制展开研究，本文采用结构方程模型和bootstrap方法对数据进行分析和检验。主要研究发现：系统质量、毒性行为、信息过载、社交过载和主观规范在使用倦怠、不满意度及两者组成的“中介链”的中介作用下对MOBA类网络游戏用户的中辍意愿产生影响，其中社交过载和主观规范在不同的路径中存在不显著的情况。

ABSTRACT. 本文采用文本分析和问卷调查方法，通过对两类代表性土味视频内容及评论的研究，揭示土味视频内容和负面评论的情感强度对用户粘性行为的影响关系。主要研究发现是：土味视频中的土味语言符号和非语言符号会对用户粘性行为存在正向影响；负面评论的情感强度会对用户粘性行为存在负向影响；视频类型在土味视频内容和负面评论情感强度对用户粘性行为中存在调节作用。

ABSTRACT. 针对区域间优质医疗资源分布不均，区域医疗服务质量可能存在较大差异的特点，进行不同地区医疗服务质量影响因素差异分析具有重要的意义。为了探究我国不同经济水平地区医疗服务质量影响因素存在的差异，以经济水平较为悬殊的北京地区和西部地区的三甲医院为研究对象，基于患者的在线医疗评论，利用改进的BTM主题模型进行评论挖掘，基于主题挖掘结果和SERVQUAL模型进行三甲医院医疗服务质量影响因素识别。利用调查问卷和因子分析法修正并建立医疗服务质量影响因素指标体系，利用相似度算法计算两地区患者对各因素的关注度，从区域层次出发，利用卡方检验等统计分析方法，详细对比北京地区和西部地区三甲医院患者对医疗服务质量影响因素的关注度、好评率和服务满意度，并针对北京地区和西部地区医疗服务质量影响因素的差异提出相关建议，为政府部门合理调配医疗资源、出台相关政策、医疗机构提升医疗服务质量提供支持和帮助。

ABSTRACT. 疫情防控常态化使得老年人缺乏陪伴与关怀，短视频迅速融入老年人的生活。然而老年人对时间概念认知的特殊意义如何影响其短视频持续使用行为仍缺乏关注。本文基于社会情绪选择理论和沉浸理论，将开放未来时间观、现在享乐主义时间观、沉浸体验、情感承诺、短视频类型等作为研究变量，构建老年人持续使用短视频APP的概念模型，以期丰富老年人信息系统使用决策和成功老龄化领域相关研究，为老年人信息系统行为提供理论支持和战略参考。

ABSTRACT. 本文基于信号理论，从医生自我价值体现信息角度出发，揭示在线医疗平台上医生个人描述性信息丰富度和擅长信息特征对用户决策的影响。主要研究发现是：描述性信息包括医术水平、教育经历和社会影响力三个方面，不同描述性信息特征变量对患者决策影响不同，用户在选择医生时更多关注医术水平和社会影响力；擅长信息总量正向影响用户择医，进而用户在择医时会关注医生的擅长信息覆盖度，而医生擅长信息颗粒度对患者择医无明显影响。

ABSTRACT. The COVID-19 pandemic is wreaking havoc globally, along with a pandemic of disinformation and misinformation. Fact-checking about the pandemic on social networking sites (SNSs) is more important than ever. However, few studies exist on the factors of fact-checking about the pandemic. Meanwhile, previous literature has implied gender differences in using information. The purpose of this study is to explore the factors related to fact-checking about the pandemic and the moderating role of gender on the impact of the factors. Based on social cognitive theory and social role theory, we established a research model to understand the factors of fact-checking on the COVID-19 pandemic and the role of gender. The analysis results show that personal factors (self-efficacy and outcome expectancy) and environmental factors (perceived skepticism and perceived ambiguity) have significant impacts on fact-checking about the COVID-19 pandemic. Meanwhile, our results reveal that gender has a moderating effect.

ABSTRACT. 本文基于感知价值与感知风险理论，探讨在线医美平台金钱或非金钱服务承诺对消费者购买意愿的影响及消费者冒险倾向、医美企业在线评分在该机制中的调节作用。通过2×2×2组间实验验证假设，研究发现：金钱与非金钱服务承诺对消费者购买意愿有不同程度的影响，相比于金钱补偿，非金钱服务承诺（即，免费术后修复）能显著提高消费者对医美服务的购买意愿；当冒险倾向或企业在线评分越低时，非金钱服务承诺对消费者购买意愿的正向影响更强。

ABSTRACT. The rapid popularity of live streaming has driven the emergence of a new business model, known as live streaming commerce (LSC). With the highly interactive feature, social interactions in LSC play an important role in affecting viewer purchase behaviors. In this paper, we collect a rich live streaming dataset and identify two types of social interactions behind danmaku, the real-time commenting that scrolls on the screen. We find that social interactions in LSC could be classified as transaction-oriented and relationship-oriented categories. Our paper shows that there exists a curvilinear relationship between relationship-oriented social interaction and viewer purchase behaviors in LSC. Based on our empirical analysis, to achieve higher sales in LSC, broadcasters should make the relationship-oriented social interactions accounting for a two-thirds share. Our paper also illustrates the different effects of social interactions on viewer purchase behaviors.

ABSTRACT. 本文使用大众点评（dianping.com）平台数据，通过双重差分模型揭示了用户生成内容平台中商户进入推荐榜单对商户的潜在影响。研究结果发现，进入推荐榜单给商户带来一定的积极影响：消费者更加愿意分享自己在该商户下的消费体验，该商户评论量显著提升；然而，由于用户初始期望与感知绩效之间的差异，商户的总评分显著下降；用户对于产品的取向更加单一，导致了人均消费额的下降。本文扩展了上榜效应的相关研究并通过分析上榜效应丰富了期望确认理论相关的文献。

ABSTRACT. 如何实时分析中国情境中社交网路上品牌的利益相关者的感受、意见、态度与想法，并以细致、量化的方式进行监控？本文以服装品牌为例针对国内最常用的社交媒体进行实时的评论爬取，提出了品牌声誉概念化框架和多级声誉感知因素，并通过文本挖掘的方式探索社交媒体所谈及的声誉因素，再构建评价体系，设计了一个品牌声誉实时评价系统。该系统可以真实、及时反映社交媒体平台中品牌声誉随时间变化情况，并分析各个因素之间的内在联系，探索利益相关者对品牌的感知过程，发现“品牌感知”与“关系”这两个声誉影响因素与公司股价业绩相关性，声誉评价数值具有实际金融意义。

ABSTRACT. AI technology empowers AI agents to recognize and convey emotion and social cues, creating frequent interaction between consumers and AI agents, and an emerging human-AI relationship emerges. To advance our understanding of this human-AI relationship, this study proposes that perceiving a mental state is positively associated with the human-AI relationship and investigates the underlying mechanisms. Specifically, a survey was conducted to test the hypotheses. The results show that two dimensions of empathy are positively associated with harmonious human-AI relationships through elevated perceived anthropomorphism and psychological empowerment. Furthermore, this study verified interaction effects, comparative effects, and service stages differences. Finally, theoretical contributions and practical implications are discussed.

ABSTRACT. 本文采用行为实验的方法，以认知信任理论为基础，探索图片评论的数字化加工程度对消费者购买意愿是否会产生影响，以及该影响如何发生。我们的研究发现：图片评论的数字化加工程度对消费者购买意愿的影响呈倒U型，消费者的认知信任随着数字化加工程度的增加而降低，且认知信任在数字化加工程度影响消费者购买意愿的过程中起中介作用。

ABSTRACT. 本文采用问卷调查的方法，揭示了政府助农直播对消费者购买意愿的影响机制。结果发现：官员主播的感知亲民性和身份权威感、直播所售农产品的区域品牌特性、直播场景的互动性、生产场景性等能通过消费者的认同、内化机制显著影响购买意愿。因此，政府助农直播应通过借助官员身份、放下官员“架子”、提供高品质、具有区域特色的农产品并借助真实的生产场景进行直播、加强与消费者互动来增强消费者的认同和内化，进而提升消费者购买意愿。

ABSTRACT. 本文首先讨论了体验型产品在线评论的不一致及其与商家信息呈现的不一致对消费者购买决策的影响作用，进而基于远程临场感理论，采用生动性、交互性、丰富性三个变量刻画AR技术特征，并探讨AR技术对消费者购买决策的影响作用。本文采用京东1000多个美妆产品的43万条评论验证研究模型，验证了在线评论的不一致性、在线评论与商家信息呈现的不一致性对消费者决策的消极影响，以及AR试妆技术对消费者决策的积极影响和调节效应。

ABSTRACT. 本文采用数据科学、设计科学研究方法，对新冠疫情背景下国家税收收入预测进行再思考与再设计，利用人工和自动参数选择的综合选参法对税收影响因素进行分析，提出一种基于LSTM-BP深度神经网络模型的多因素时间序列预测方法。实验证明，本文的预测方法综合了LSTM网络与BP网络两者的优点，在国家税收收入预测任务方面比其它基线模型有更好的预测精度，并从多因素的角度增强了预测结果的可解释性。

ABSTRACT. This paper constructs a research theoretical model based on the social identity theory to examine the effects of in-formation disclosure and emotional disclosure of posts on the number of replies received to a post in online health communities for people with depression. There are several findings in our research. First, information disclosure negatively impacts the number of replies. Second, when posts show higher levels of negative emotion, they receive higher replies, and the degree of suicidal tendencies positively impacts the number of replies, while positively mod-erating the effect of information disclosure on the number of replies.

ABSTRACT. The market performance of mobile apps is very important to developers, but it is difficult to maintain or improve in the highly competitive market. An essential method to achieve the competitive advantage is iterative and incremental to update the apps. We propose a dynamic approach to capture the update strategy of mobile apps. Specifically, Hidden Markov Models are built to examine the relationships among update strategy, user satisfaction, and app performance. Theoretically, we propose that the update strategy determines the transmission between various levels of user satisfaction, which is the hidden states in HMM. Then, the transmission of user satisfaction results in app performance. The critical contribution of our study is that the dynamic process of mobile apps updates is theoretically and empirically examined. In this RIP, we propose the research design and discuss the implications.

ABSTRACT. The emergence of the epidemic has fueled a series of medical chatbots that have sprung up, and although domestic healthcare-oriented chatbot practices are currently flourishing, there are problems of low patient trust. Compared with foreign countries, research on the application and effects of chatbots is far from adequate, and there is a lack of systematic review of the development and application of intelligent chatbots in the mental health field. This study identifies the status by analyzing the application and category distribution of chatbots in mental health, with the aim of providing references and insights to the practice and related research of chatbots in mental health applications in China.

ABSTRACT. 传统的问答社区专家推荐方法主要将问题的标题和内容看作一个整体来对专家的整体兴趣进行建模，忽略了专家会仅根据标题初步筛选感兴趣问题的事实。基于涉入理论，本文分析了问题的标题和内容在专家点击标题、回答问题决策中的不同作用，并提出了新的同时考虑点击和回答预测任务的专家推荐方法。本文利用心理健康问答平台“壹心理”的真实数据进行了实验，验证了本文方法针对传统的未分离标题和内容的各种推荐方法的优越性。

ABSTRACT. 本文采用功能性近红外光谱技术，探究安全警告威慑、安全知识引导、安全警告威慑+安全知识引导组合这三种干预对网络钓鱼敏感性的触发机理，同时探究认知加工和信任的中介作用以及情景的调节作用。通过认知神经实验结果发现，安全警告威慑对网络钓鱼敏感性无显著触发效果，安全知识引导、安全警告威慑+安全知识引导组合均显著触发网络钓鱼敏感性；认知加工在安全干预对网络钓鱼敏感性触发中起中介作用，而信任的中介效应不显著；情境在安全干预类型和网络钓鱼敏感性关系中存在调节作用。

ABSTRACT. 数字经济背景下的隐私威胁事件频繁发生，用户对隐私威胁事件进行评估后会采取隐私保护行为，探究威胁评估对用户隐私保护行为意愿的影响机理，有助于企业制定合理的隐私管理决策进而营造健康的企业数字生态。本研究基于保护动机理论并聚焦威胁评估，探究影响威胁评估对用户隐私保护行为意愿的两组关键因素：情感和情境的信息敏感度。引入“信息隐私焦虑”为情感中介变量，构建威胁评估对用户隐私保护行为意愿的影响机理模型，根据前景理论，探究情境的信息敏感度对该机理模型的调节作用。运用SEM-PLS对收集的金融情境183份与电子商务情境200份有效数据进行实证分析。研究结果表明：信息隐私焦虑是影响用户产生隐私保护行为意愿的关键情感因素，信息隐私焦虑在感知威胁和隐私保护行为意愿之间起部分中介作用；情境的信息敏感度正向调节信息隐私焦虑与隐私保护行为意愿的正向关系；情境的信息敏感度仅对感知脆弱性与感知威胁的正向关系具有调节作用，而对感知严重性与感知威胁的正向关系没有调节作用。本研究补充和发展了现有的保护动机理论，对企业采取合适的管理措施以减少用户消极的隐私保护行为提供了理论指导。

ABSTRACT. Based on Theory of Business Model of the Stakeholder’s Deal Structure , This study aims at the needs of rural culture-oriented tourism destinations, and applying grounded theory, conducts case studies on three typical OTA platforms (Ctrip, Qunar, Mahive), three enterprises of supply chain in rural culture-oriented tourism industry (OCT, Xinhua International Travel Service, Dayou), and three rural culture-oriented tourism destinations (Wudang Mountain, Jiangwan town, Liuzhong Township). Seeking new interested parties, enhancing the resource capacity and transaction value of the main trading parties of the culture-oriented tourism platform, optimizing the transaction structure, and aiming at the limitations of the current OTA business model, this paper puts forward the business model of "third party+"intelligent service platform of rural culture-oriented tourism.

ABSTRACT. In the context of decentralized communication, how to promote the public to understand and adopt authoritative content is a topic of common concern to government decision-making departments and academia. Based on the meaning construction perspective, this paper takes 31 provincial health-related government microblogs in China as the research object, and uses the fuzzy set qualitative comparative analysis (fsqca) to gain insight into the combined interaction effect of the constructive communication characteristics presented by health information and the public decision-making environmental conditions on the public information adoption behavior. The study found that: a single condition does not constitute a necessary condition for a high level of public health information adoption, but the proper use of a combination of constructive communication features plays a more universal role in improving public information adoption behavior; Among the eight equivalent driving paths formed by the combination of constructive communication characteristics and decision-making environmental conditions, three types of government constructive communication adaptation modes are further summarized. The empirical results have important reference significance for improving the communication ability of local government new media.

ABSTRACT. 基于交易双向契合的理论，将零售创新解构为零售企业与消费者交易契合及其过程的创新，采取单案例研究方法，构建了数字化驱动零售创新演化机制模型。研究发现：数字化驱动零售创新遵循“消费者数据化参与—零售企业数字化能力构建—交易契合创新”的微观实现路径；结合突破时空约束和降低信息阻力的两个演化维度，这经历了单个门店、连锁门店和全渠道的演化过程，其中消费者数据化参与不断深化、零售企业数字化能力持续升级，从而促使交易契合向满足消费者个性化场景需求的方向演化。本研究补充了数字化情境下零售创新微观实现及其演化机制的研究缺口，对零售企业数字化创新路径或转型过程具有研究启示。

ABSTRACT. In order to fully realize the business value of cloud computing and achieve organizational objectives, it is extremely important to clarify the impact of different environmental characteristics on the ambidextrous innovation value of cloud computing. However, the existing literatures do not fully investigate the moderating role of environmental factors. This study conducts propensity score matching and difference-in-differences-in-differences analysis on the secondary hand performance panel data of 118 pairs of listed companies to investigated the moderating effect of the three environmental characteristics, namely dynamism, munificence and complexity, on the ambidextrous innovation value of cloud computing. The results show that cloud computing significantly promotes exploitative innovation in less dynamic environment and explorative innovation in high complex environments. Nevertheless, munificence explains a less significant proportion of the variance in ambidextrous innovation performance. This research creatively and empirically reveals the realization regularity of cloud computing ambidextrous innovation value under different environmental characteristics which provides theoretical and empirical implications for managers to carry out cloud based operation.

ABSTRACT. On mobile apps platforms, management responses enable developers to spotlight customer feedback and then react accordingly. Aiming to understand management responses in digital innovation, we develop a theoretical model of attention-allocation that addresses an underexplored but essential question: In a digital system, how is spotlighted customer feedback absorbed into the following innovation, and what role do management responses play in this absorption? To answer the questions, we first identify the spotlighted customer feedback, relevant management responses, and the subsequent mobile apps updates on the Apple iOS platform. Secondly, text mining algorithms are developed to analyze the feedback and responses, and then measure stimulus-driven attention, goal-oriented attention, and the knowledge absorption in subsequent updates. The empirical results reveal that responding to maintenance and functional customer feedback improves the subsequent knowledge absorption in mobile updates. Additionally, absorbing knowledge from customer feedback is also improved if developers customize and post actionable response.

ABSTRACT. With the rapid development of online music platform, music socialization has become one of the mainstream online social forms. For the “effect of people with the same preference gathering together” characteristic of playlist comment section, this study adopts the uses and gratifications theory, technology acceptance model to construct a theoretical framework for the influence of users’ virtual social needs on the playlists’ selection behavior. In Study 1, Python technology was used to capture the playlist comment data. Based on the uses and gratifications approach, and the qualitative analysis method of grounded theory, five dimensions of users’ virtual social needs were obtained, namely, the memory resonance, emotional expression, entertainment, interaction and self-expression needs. Study 2 established an econometric regression model to verify the causal relationship between users’ virtual social needs and playlist selection behavior, using the number of plays, shares, and favorites as the evaluation indicators. The study found that users’ needs for entertainment, self-expression and emotional expression positively influence playlist sharing behavior; while the favoriting and playing behaviors are mainly positively affected by memory resonance and entertainment needs. Study 3 verified the causal mechanism between users’ social needs and the willingness to select playlists by constructing a structural equation model (SEM). Results showed that, different virtual social needs would further influence users’ willingness to play, share, and collect playlists by affecting the perceived usefulness, perceived ease of use, and perceived enjoyment, respectively. The purpose of this study is to reveal the users’ virtual social needs, behavioral patterns and influencing factors of music platforms, and to provide theoretical guidance for the recommendation strategy of high-quality playlists on music platforms.

ABSTRACT. O2O（Online to offline）平台将本地服务企业与消费者相连接，深刻地影响了本地服务企业的经营与绩效。在线消费者构成，本地企业的服务相关决策与平台相关决策将共同影响O2O平台上本地企业的绩效。基于平台生态系统视角，引入组态思维，运用模糊集定性比较分析（Fuzzy-set qualitative comparative analysis, fsQCA）方法探索影响O2O平台内本地服务企业绩效的因素组态。以影院为样本，采用比较静态的分析方法，考察在O2O平台的成长期与稳定期，绩效因素组态发生了什么改变。研究表明，高票价和高线上销售占比条件变量出现在影院所有高绩效组态中。在平台成长期，高排座率、低排片多样性和高自建平台具有替代性；而平台进入稳定期，排片和自建平台成为影院获取差异化竞争优势的关键。

ABSTRACT. As urban transportation commuting causes serious pollution problems, green commuting has aroused increased interest in academia. However, there is a lack of understanding of whether or not wealthy residents prefer green commuting and when residents choose green commuting. This study investigates how income (i.e., housing prices) affects residents’ green commuting choices. We integrate a novel data set to generate a unique dependent variable to evaluate residents’ green commuting rate, by combing two large real world transportation data sets. Our empirical analyses indicate a negative relationship between housing prices and green commuting. Additionally, residents’ green commuting choices differ across different commuting purposes. Specifically, the green commuting rate decreases for working, hospital visiting and hedonic purposes. This study contributes to the literature on urban traffic management by using real-world data to examine the factors affecting green commuting, while most of the research in this field is conducted using survey methods.

ABSTRACT. 本文采用经济学建模方法，探究在双寡头竞争市场下，企业利用从数据供应商购买的数据来为用户提供个性化 产品的策略，给出了企业的数据购买策略，产品定价策略以及市场均衡。研究的结果表明:在均衡时，数据供应商会 将数据同时卖给两家企业。但相互竞争的企业也都会采用个性化定价的方式，最终导致了基于数据的产品个性化加剧 企业的价格竞争，降低企业利润，但增加了消费者剩余和社会福利。

ABSTRACT. 数据交易是数据要素市场化配置的核心环节。针对数据需求者不能事前掌握被交易数据产品的质量、完整性和效用（即存在感知价值不确定），探究数据产品抽样策略，以提升数据需求者购买意愿是数据交易平台可持续运营过程中亟需解决的关键问题。本文构建双寡头竞争博弈模型，分别探究两个平台均不采取数据抽样策略、一个平台采取数据抽样策略和两个平台都采取数据抽样策略三种情景下的最优数据产品质量和交易价格。依据数据需求者对数据产品的需求偏好，将数据交易市场分为以数据匹配为主导和以数据质量为主导的两类数据交易市场，改进传统的Hotelling模型刻画数据抽样策略对数据需求者效用的影响，分析均衡状态下数据产品质量、平台的收益与平台数据抽样策略间关系。研究结论为数据交易平台采取数据抽样策略提供了理论指导，同时对改善平台的活跃度和数据交易量提供了新的实践路径。

ABSTRACT. 移动健身技术已经广泛应用于人们的运动行为。移动健身技术为用户提供的数字化运动信息，能否以及如何影响用户的持续健身行为，已有研究还少有涉及。基于自我调节理论，本研究探索移动健身技术对用户持续健身行为的影响机制，考虑用户健康赋权和自我提升意愿的中介机制，及锻炼目标清晰度和自我控制的调节作用，研究结果对已有移动健身研究和实践具有重要启示。

ABSTRACT. 摘要：［目的/意义］非遗信息搜寻作为短视频生态重要的一环，挖掘其用户信息搜寻行为的影响要素及其作用机理，有助于增强非物质文化遗产活力，提升非物质文化遗产的现代化传播力度。［方法/过程］本研究以刺激-机体-反应理论和传播生态学理论为基础，通过调查收集数据，利用层次回归分析法检验变量间的主效应和中介效应关系；进而运用模糊集定性比较分析方法探索上述因素间的组合效应。［结果/结论］研究发现内容层中短视频权威人物和社会层中的临界质量是影响用户搜寻意愿的主要因素。其中，内容层的权威人物是短视频满意度产生的必要要素；短视频满意度在短视频内容呈现与用户信息搜寻意愿间起到完全中介作用，在短视频系统质量、短视频权威人物、临界质量、虚拟社交距离与用户信息搜寻意愿之间存在部分中介作用。本研究丰富了短视频情境下非遗信息搜寻行为的理论基础，同时对非遗信息的制作和传播提供了相应建议。

ABSTRACT. Emerging research reveals the relationship between the Internet and public’s attitude towards healthcare. Some holds that Internet will improve their attitudes while others believe that it will deepen their misperceptions about healthcare. However, there is a lack of research that further disaggregates public attitudes into different dimensions and explores the different impacts of Internet use and Internet involvement. To answer this inconsistency, this study distinguished these two and divided the attitude toward healthcare into three aspects: doctor trust, medical institution satisfaction and medical problem perception. And this study used propensity score matching to analyze the effects of Internet on residents’ attitude by adopting the data from the China Family Panel study in 2018. The results show that Internet use only contributes to the rise of medical problem perception level. And Internet involvement increases Chinese residents’ doctor trust, but also ruined their medical institution satisfaction and exaggerated medical problem perception.

ABSTRACT. 数智化浪潮为人文学者数字学术研究和实践注入了新的动力，但与此同时学者们在与数字内容和技术交互中也面临数字囤积困境。本文采用半结构化访谈、关键事件技术法和主题分析法对人文学者的数字囤积行为进行了探究，包括对人文学者数字囤积行为的概念化解析、行为模式的动态演化特征及囤积者类型的识别、数字囤积情境因素的挖掘，并在此基础上构建多元要素和多情境方向发展的数字囤积行为形成机理模型。最后，从系统设计和技术赋能的实践视角出发，为提升信息系统对于解决人文学者数字囤积困境的赋能优势提供了参考。

ABSTRACT. [Purpose / significance] This research aims to explore the influence mechanism of privacy disclosure intention of young elderly under the background of aging-friendly retrofit, and provide suggestions for further promoting reform and protecting the rights of the young elderly. [Method / process] Combined with privacy calculus theory, PLS-SEM was used to analyze the influencing factors of young elderly's perceived benefits and perceived risks, and to clarify its mechanism on privacy disclosure intention. [Result / Conclusion] The results show that personalization, procedural fairness and information control positively affect the perceived benefits of the young elderly, and then promote their intention to disclose privacy. Subjective norms and information sensitivity positively affect the perceived risks of the young elderly, and then inhibit their intention to disclose privacy. In addition, health information overload plays a negative regulatory role in the impact of perceived risk on privacy disclosure.

ABSTRACT. 智能问诊聊天机器人提供了一种专业化、即时响应、低成本的健康信息咨询工具。本文从功能、社会、用户动机三个特征维度，构建在线智能问诊聊天机器人用户采纳前因组配模型。采用fsQCA和访谈相结合的混合研究方法，通过收集347名用户的调查问卷数据进行模糊集定性比较分析，并与后续12位半结构化访谈结果相结合。研究结果共发现了五种因果组配方案，感知社会存在是核心条件。本研究为健康服务机器人的开发者和制造商提供了实践指导。

ABSTRACT. 传统线下就医行为中，人们往往盲目追逐大城市大医院，造成优质医疗资源紧缺而普通医疗资源被闲置的窘境。为了解决这一问题，本文采用行为学实验的方法，依据精细加工可能性模型探究不同类型用户在线上选择医生的行为决策的理论机制与实际表现。结果表明不同类型的用户在线上选择医生时所偏好的医生类型不同，这一结果使得我们针对不同类型的用户推荐不同的医生，发挥在线问诊平台分医导流，合理配置医疗资源的作用成为了可能。

ABSTRACT. Background With the flourishment of e-health in recent years, e-consultations provided by doctors in Online Healthcare Communities (OHCs) are becoming more and more popular with the public in China. Meanwhile, proactive behaviors of doctors attribute to the success of OHCs, and they are often unsure how to behave proactively to attract patients, with limited research focusing on this. This study aims to explore how proactive behaviors of doctors affect their individual e-consultation service quantity in OHCs. Methods Data in our empirical analysis was collected from one representative OHC in China, referring to 3170 doctors in 865 medical teams. Hierarchical linear regression was conducted to examine the effects of doctors’ proactive behaviors at different levels on individual performance. Herein, proactive behaviors of doctors in OHCs mainly included response speed, pricing and related behaviors in the context of online medical teams. Results This study found that proactive behaviors at different level had significant impacts on individual Service Quantity (SQ) of e-consultations in OHCs. Further, positive effects could be verified in most conditions that were congruent with the logic of common sense, and the interaction terms demonstrated complex influences. Specifically, at the cross level, the switching cost weakened the relationship between price at individual level and individual SQ (β=-0.001, p<0.01), while strengthened the relationship between the response speed (β=0.110, p<0.01) at individual level and individual SQ. At the team level, the response speed in online medical teams positively moderated the effects of both price (β=0.438, p<0.01) and response speed (β=33.678, p<0.01) at individual level on individual SQ. In summary, proactive behaviors at different levels and their interaction effect play key roles in individual SQ. Conclusion Results of empirical analyses enable healthcare providers to adjust their behaviors proactively. In brief, our research provides healthcare practitioners with an improved understanding of the impacts of doctors’ proactive behaviors, and ultimately improve the efficiency and provision of delivered healthcare services in OHCs.

ABSTRACT. Physicians are the primary providers of online medical services, and their participation is critical to the vitality and sustainability of the online healthcare community. Drawing upon the affordances theory and job demands-resources model, we integrate required online effort and platform affordances in online work settings and propose a dual-path model of online retention for physicians. The expected findings are that physician, cognitive, and emotional labor discourage online retention of physicians, egocentric affordance, online collaboration affordances, and social presence affordances contribute to it, and job crafting mediates these effects. The expected findings may extend the understanding of the theoretical motivation behind continuance behavior and help platform managers to motivate physicians’ continued participation on the platforms.

ABSTRACT. 摘要：在线健康是结合了互联网和卫生服务行业的新兴领域，近年来，由于科技的发展和疫情等情况的出现，该领域的研究热点也在不断变化。本研究利用CiteSpace可视化工具对Web of Science数据库在2016-2020年间收录的在线健康主题的文献进行可视化分析，绘制出知识图谱，揭示该领域的研究热点和前沿，探讨未来发展趋势，为后续在线健康领域的相关研究提供参考。研究发现，健康焦虑、癌症、数字健康、初级医疗以及COVID-19等是近年在线健康领域的新兴研究点。

ABSTRACT. Online health consultation, as one of the most popular and fast-growing online health services, can relieve the pressure of offline medical care and is widely valued. Paid health consultation in Q&A platforms is a novel form of online health service where users other than physicians can provide health consultation services as respondents. In this study, we focus on the effect of respondents’ username-physician fit on their online consultation using a unique panel dataset of 254 respondents from a large online Q&A platform in China. In addition, the moderating role of online reputation and online effort is explored. The results show that the respondent's username-physician fit in the Q&A platform has a negative impact on the online consultation, that is, there is a negative halo effect. Furthermore, we find that online reputation and online effort have a moderating effect on this negative effect.

ABSTRACT. Information technology has shown great potential in healthcare, especially in chronic disease management. In this paper, we leverage rational addiction model to investigate patients’ behavioral dynamic relationship in the context of IT-enabled diabetes self-management. We firstly build an analytical model to capture the impact of patients’ online learning behavior on their health outcome and the patients’ decision making about their self-management behaviors, then utilize a real-world data to empirically test the hypotheses. The GMM estimator of dynamic panel data model demonstrates the rational forward-looking behaviors generally exist in patients with diabetes, but they exhibit heterogeneous patterns depending on their demographics such as age, disease duration and gender. We contribute to the literature of IT-enabled chronic disease management, and also extend the application context of rational addiction model.

ABSTRACT. 自20世纪90年代以来，随着互联网日益普及，在线医疗已成为近年来必不可少的信息通信技术应用。而随着在线医疗产业界的蓬勃发展，国内外众多学者对在线医疗进行了大量的研究，形成了一个鲜明的研究现象。本文献综述从1）在线医疗信息系统采纳主题、2）信息系统顶刊相关发文趋势、3）用户生成内容相关研究三个角度回顾了本领域学术成果发展脉络，揭示了信息系统学科在该领域的未来研究方向。

ABSTRACT. Contents and information behaviors as the main vector of micro-blog can be analyzed to provide insights into what, why, and how contents influence information behaviors in crisis situation, which help crisis management understand group users’ behaviors and the characteristics of sociocultural systems. In our research, we examine whether the affective dimensions and non-emotion dimensions (using Cox proportion hazard model extract features) of micro-blog contents associated with diffusion behavior and non-diffusion behavior in terms of quantity in the social media context of earthquake event. Results show that emotion contents have more significant influence on diffusion behavior than non-diffusion behavior, while non-emotion contents have more significance effect on non-diffusion behavior than diffusion behavior. The results give implications for both practitioners and scholars who are interested in further studies on crisis management and information behaviors research in the social media context.

ABSTRACT. 在直播电商行业快速发展却伴随着高退货率这一痛点问题的现实背景下，本文采用ABC态度模型，通过问卷调查方法，探究直播电商中消费者退货责任归因对其后续在直播间购买意愿的影响和矛盾态度的中介作用。结果表明：个人归因负向影响消费者对主播的矛盾态度并正向影响重购意愿；商家和主播归因正向影响消费者对主播的矛盾态度并负向影响重购意愿。消费者对主播的矛盾态度在退货责任归因与重购意愿间起到中介作用并负向影响重购意愿。

ABSTRACT. With the development of e-commerce platforms, more and more platforms pay attention to social information, especially action-based social information (i.e., peer consumer purchase). However, different e-commerce platforms expose different peer consumer purchases. The peers are not online but face-to-face. We introduce offline relationship into the online decision-making process of consumers and we also pay attention to the relative impact of peer consumer purchase. We conducted an experiment to manipulate the degrees of peer consumer purchases as well as utilitarian and hedonic products. We get the following result: 1. exposing the number of peer consumer is the most suitable way to improve consumers’ purchase intention; 2. product type plays an important moderating role in consumer purchase decisions. This study contributes to the existing literature on social information exposure about historical sales. And the results provide practical suggestions for e-commerce platforms on how to expose peer consumer purchases for different product types.

ABSTRACT. 本文主要基于新冠疫情数据和在线旅游评论数据，探究新冠疫情对游客旅游情感的影响效应。首先，基于现有理论建立新冠疫情与旅游情感的假设关系，再通过PAD（Pleasure-Arousal-Dominance）三维情感分析模型提取旅游评论中的游客PAD情感，接着建立实证计量模型，最后运用双重差分方法估计模型，并且进行一系列稳健性检验后，以此得出最终的实证结果。研究结果发现，新冠疫情已经给游客情感带来显著影响效应，同时检验了旅游景区类型和旅游景区等级的调节效应。

ABSTRACT. 随着在线社交推介逐步成为社交电商营销主要模式之一，多任务社交推荐系统设计对驱动用户参与社交分享的影响，成为电子商务界研究的重要议题。本文以社会交换理论为基础，采用实验分析方法，重点探究了多任务社交推介情境下，商家推介系统信息框架设计和分享机制对用户参与意愿的影响。研究发现：在多任务社交推介场景下，相对于任务导向信息框架设计，奖励导向的信息框架设计更有利于用户参与推介活动。

ABSTRACT. 本文依据目标梯度效应理论，针对包含折扣和满减两种促销形式的多重折扣广告，实证研究多重折扣中满减阈值效应。研究发现，多重折扣中折扣价格和满减阈值的相对距离越小时，消费者越愿意点击广告。进一步分析发现，对于低知名度品牌及促销力度确定时，相对距离越小，消费者点击意愿越强。而对于高知名度品牌则情况相反。并且促销力度不确定时，这一影响消失。此外多重折扣力度越大，相对距离对消费者点击行为的负向影响越强。

ABSTRACT. 随着电子商务的飞速发展，在线商品评论成为影响消费者购买决策和产品销量的重要因素。然而，消费者对在线商品评论的依赖导致在线商品虚假评论的不断涌现，电商平台和商家通过操纵评论进行恶意竞争，在损害消费者和商家相关利益的同时也扰乱了电子商务营商环境。本文以在线商品评论的文本数据为主要研究对象，基于说服性知识理论探究评论感知真实性对消费者决策的影响。一方面，本文结合数据分析和实证模型探究评论感知真实性对消费者决策的影响。另一方面，本文通过文本数据分析，探究影响在线商品评论感知真实性的评论文本表现形式。本文构建系统广义矩估计模型对在线商品评论感知真实性对消费者决策的影响进行实证检验。同时，本文构建fractional logit模型探究影响在线商品评论感知真实性的评论文本表现形式。研究结果表明：在线商品评论的感知真实性通过调节消费者对评论信息的处理过程从而影响消费者决策。评论的参考性、语境嵌入、内容细节和论点结构化是影响在线商品评论感知真实性的主要表现形式。研究结果为商家和平台有效治理在线商品虚假评论提供管理启示。

ABSTRACT. 制造业采购经理人指数（Purchasing Managers’ Index ，PMI）是反映国家经济运行情况的重要指标，而传统预测模型对该类时序数据预测精度不高。针对制造业PMI指数的非线性、波动性和数据量少的特点，提出一种基于一维离散小波变换进行数据预处理的组合模型，时序数据经过小波变换，由整合移动平均自回归-广义自回归条件异方差模型（Auto Regressive Moving Average-Generalized Autoregressive Conditional Heteroscedasticity，ARMA-GARCH）处理稳态低频数据，门控循环单元（gated recurrent unit，GRU）处理波动性强的高频数据，将各频段预测结果进行融合得到最终预测结果。为验证模型有效性，选取一定数据量的PMI指数进行实验，结果表明，与其他常见模型对比，本文构建的组合模型具有较好的预测精度与性能，平均绝对误差（MAE）、均方根误差（RMSE）、平均绝对百分比误差（MAPE）分别达到0.00329、0.004162，0.65%。

ABSTRACT. 作为一类新兴的隐私计算方法，联邦学习技术联邦学习实现了“数据不动而算法动”，从而为解决“数据孤岛” 提供了新的技术解决途径。然而，每个企业加入数据联邦的决策与其收益分享方式直接相关，如果不能通过合理的收 益分享方法来为企业提供参与激励，联邦学习技术也无法成功地促使企业之间真正实现其算法目的。本文模拟了一个 企业的随机图网络并讨论了在不同收益分享机制设计下，企业之间如何形成数据联邦及其算法收益。本文的核心结论 是，三种联邦收益分享机制都有可能成为最优机制，这主要取决于联邦学习算法利用数据能力和联邦企业网络规模大 小的影响。

ABSTRACT. Traditional machine learning methods have been widely applied to identify the determinants of depression in the elderly, while the lack of explainability of machine learning has been a long-standing limitation. Based on the 2018 China Health and Retirement Longitudinal Study, we combine the SHAP interpretation method with the LightGBM machine learning model to study the influencing factors of depression in the elderly. Results show that health status, socioeconomic status, family support, family structure, and social interaction are of key importance in identifying depression in the elderly. This study helps to demonstrate the effectiveness of SHAP in ranking influencing factors of depression in the elderly, and the results provide valuable practical insights in the prevention of depression in elderly.

ABSTRACT. 为了应对医疗小样本问题，精准辅助医护人员开展传染性分类、风险预测管理。本研究基于计算设计科学范式设计、实现和评价一个基于小样本数据增强的重症肺炎患者传染性分类、风险因素识别与预测系统，以帮助医护人员管理重症肺炎患者。该系统包括四个组件：基于小样本学习的数据增强、基于可解释性的 LGBM 传染性区分模型、基于 Cox 回归的风险因素识别与评估模型。通过小样本学习获得更多高质量的研究案例，在此基础上利用传染性区分模型、风险因素识别与评估模型为医生提供预诊与风险预后等智慧服务。本研究不仅对开展小样本数据增强以及分类研究具有重要的理论意义，而且对于目前的新冠防治工作具有现实意义。

ABSTRACT. 目前农产品供应链存在着各主体信息化程度不一致、信息共享程度不高导致“牛鞭效应”等问题。为促进农产品供应链各主体间信息与数据的高度共享，本文基于区块链理论分析了区块链与农产品供应链信息共享的耦合性，从农产品供应链多源信息的采集、处理和共享方面构建基于区块链的农产品供应链多源信息共享模型。研究农产品供应链的多源信息共享问题有着重要的实践意义，本研究的结论为未来相关供应链知识管理研究提供新的视角与思路。

ABSTRACT. 数字时代下的移动应用市场存在诸多不确定性，《个人信息保护法》作为我国第一部个人信息保护方面的专门法律，使企业面临着对消费者数据收集使用与个人信息保护之间的矛盾。为了缓解该矛盾，本研究提出将《个人信息保护法》当作缓解消费者隐私不确定的契机。为此，本研究总结归纳了《个人信息保护法》中企业需要告知消费者的隐私相关信息，理解了消费者对这些隐私相关信息的偏好，提出并设计了面向移动应用的隐私告知方法，接着通过三项研究，系统探究了消费者对App感知的隐私不确定性，以及隐私告知对消费者隐私不确定性及其App下载意愿的影响机理。研究结果表明，隐私不确定性在中国移动应用市场具有普遍性和情境依赖性，隐私告知能够有效缓解用户感知的隐私不确定性，并负向调节隐私不确定性对App下载意愿的消极影响，本研究首次将《个人信息保护法》应用到个人信息保护技术方法研究，为企业在法律制约中寻求发展机遇指明了路径。

ABSTRACT. Based on the TOE framework, this paper explores 22 enterprise intelligent customer service application examples from three aspects of technology, organization and environment, and expounds the influencing factors and path types of enterprise intelligent customer service application combined with csQCA method. This paper summarizes the condition configuration of intelligent customer service application, and puts forward two types of enterprise intelligent customer service application path supported by practical results, which are "digital transformation" and "Internet gene". The research conclusion provides ideas for the application of enterprise intelligent customer service, and also provides theoretical support for the application practice of enterprise intelligent customer service in complex situations.

ABSTRACT. 现代证券市场的交易已经逐渐被机器和算法渗透。人工智能时代，证券市场中程序化交易的交易机制与潜在风险均发生了质的变化。通过对当前的监管现状进行梳理，不难发现我国的程序化交易监管依然具有较多的局限性，传统规制方式显得力有不逮。为完善当前的程序化交易监管制度，需要更迭监管理念，在风险可控的前提下“适度发展”程序化交易；从事前、事中、事后三个维度，打造立体化的金融科技风控体系；将监管方式体系化、规范化，形成合理且具有稳定性的监管治理模式。人工智能时代，程序化交易的监管应当与时俱进，跟上技术发展的步伐。

ABSTRACT. 双寡头广告竞争是一个备受关注的市场现象。在现有文献中，Lanchester模型是最经典的双寡头广告竞争模型之一。当前，广告的效果受到广告弹性和口碑效应的影响越来越大。然而，在Lanchester模型的相关研究中，很少对这两个因素进行全面和深入的讨论。因此，本文提出了广义的Lanchester模型，综合地考虑广告弹性和口碑效应两个指标，并采用一般化的参数来描述相应的指标，而非固定的常数。基于该模型，本文构建了有限时间的双寡头广告博弈模型。接着，本文对该博弈模型的闭环均衡广告策略及其稳态进行了分析，推导出其理论性质并给出了数值解。在此基础上，本文进一步对广告弹性和口碑效应进行了参数敏感性分析。本文的模型可以更准确地描述广告竞争现象，从而为双寡头市场下的企业提供更有效的广告策略。

ABSTRACT. 随着网络的普及和发展，越来越多人在社交媒体上讨论婚姻相关的社会关注的热点话题，比如彩礼的意义及必要性。然而，少有学者研究这种讨论对参与者的婚姻信任的影响。为填补这一研究空白，我们从微博上收集了与彩礼讨论相关的帖子。基于契约理论，我们构建了研究模型和假设，并采用计量模型检验了在线彩礼必要性的讨论行为和婚姻信任之间的影响机制。此外，我们也探索了自身的固有认知和他人的社会互动对主效应的调节作用。

ABSTRACT. 面向在线健康社区需要向用户推荐既感兴趣且能改善其健康状况的干预方案。为此，本文提出了融合效用与兴趣的在线用户健康干预推荐模型，先利用注意力机制对用户兴趣进行学习，同时计算待推荐方案对其参与者的预计疗效，接着评估干预方案对目标用户的效用，最终结合方案效用和用户兴趣得出推荐结果。实验结果表明与基线推荐模型相比，本文所提出的模型不仅具有更好的推荐精度，还能有效提高所推荐干预方案对于用户健康改善的效用。

ABSTRACT. Based on the company financial reports and employee reviews on the Glassdoor platform, a new data sources, this paper employs the Word2vec model to mine the company's declared culture and employee perception culture from two types of texts in the perspective of corporate culture. On this basis, a new measurement of employee-organization fit is proposed. Then we examine the impact of employee-organization fit and employee online rating on company value. The results show that: under the new measurement method, there is a certain difference between the company's declared culture and employees' perceived culture as a whole, and the employee-organization fit of most companies is at a high level; employee-organization fit has a positive impact on company value, and it will enhance the promotion effect of employee online rating on company value. This paper makes a new attempt to measure employee-organization fit by using new data, and clarifies the value of employee online rating and employee-organization fit, which has theoretical and practical guiding significance for related research and organizational management.

ABSTRACT. 针对金融欺诈检测方法业务可解释性低，忽视欺诈个体间的关联信息特征提出融合多特征的改进Stacking金融欺诈检测模型。该模型结合有监督分层策略，从行为和关联这两个信息维度挖掘账户特征，在改进Stacking集成学习框架中实现异质特征融合，并在某互联网金融公司交易数据集上实验。证明改进Stacking集成学习框架显著提高了欺诈检测效果，避免在数据端直接融合导致的特征维度冗余问题。同时实验证明融合关联特征的欺诈检测模型在准确率、召回率等各项指标上都表现出更好预测性能。

ABSTRACT. 以往的研究发现美观度可以影响消费者和投资者的选择偏好和价值判断。然而，现有的文献一方面仅关注具体的美学元素（如图片、颜色、字体和图文比例等），缺乏对美观度这一更加全局和抽象概念的衡量；另一方面主要围绕对公司感知形象的影响，缺乏针对公司实际经济利益的定量结论。本文以2007-2019年发布社会责任报告的上市公司为研究对象，首次采用深度学习方法提取社会责任报告中的美观度指标，并考察了社会责任报告美观度与上市公司债务融资成本的经验影响关系。结果发现：上市公司的社会责任报告美观度越高，其债务融资成本越低，意味着更美观的社会责任报告能够更有效地向债权人传递企业履行社会责任和披露信息的积极信号，降低债权人对不确定性的评估，增强债权人对企业还贷的信心。社会责任报告美观度对债务融资成本的抑制作用对分析师关注度高和机构投资者持股比例高的企业更加显著。此外，进一步的分析表明，媒体关注度同样会增强社会责任报告美观度对债务融资成本的抑制作用。本研究不仅从技术增强视角探索了深度学习方法对企业社会责任报告研究的作用，也为公司披露社会责任信息和制作社会责任报告提供了有益的启示。

ABSTRACT. 开发利用数据资源是数字中国建设的重要任务,开展基于用户评论内容的数据资源应用是信息资源管理学科的前沿交叉科学问题。本文针对基于在线评论情感分析的鲜果动态定价问题，首先建立融合边缘采样和协同训练的在线评论情感分析方法，然后构建基于在线评论情感分析的鲜果动态定价模型，最后通过融合高斯回代的交替方向乘子法求解模型。研究结果表明，本文提出的算法在收敛性上有优越性，为鲜果动态定价问题提供了新策略。

ABSTRACT. The widespread use of cloud computing has brought cloud security to the forefront. The provider and the firm take joint responsibility for cloud security with cloud service models, including IaaS, PaaS, and SaaS. As an essential participant, the strategic hacker attacks the cloud system resulting in cloud data leaks. In this paper, we construct a game-theoretical model to study cloud security management, where the game players include a cloud provider, a firm, and a hacker. By comparing the case of the strategic hacker and the case of the non-strategic hacker in the model, we find that the presence of the strategic hacker has a significant impact on cloud security investments. Different levels of attack can lead to underinvestment or overinvestment by the provider and the firm. In addition, our results also show that the compensation rates in bilateral refund contracts (BRCs) are different because of the existence of the strategic hacker. Furthermore, we examine the interaction between the three parties by considering the cloud service model. The strategic hacker’s attack effort is U-shaped with the cloud service model from IaaS to PaaS to SaaS. Meanwhile, the compensation rate has an increasing trend. Based on it, the effort put in by the provider and the firm also responds with the specific cloud service model, while they have the free-riding problem due to joint responsibility. Besides, from the perspective of social welfare maximization, there are cloud security investment problems in BRC. The provider and the firm would have the dislocation investment problem to maximize their own payoffs. Finally, our paper proposes two new contract types which monitor and verify respective efforts through internal and external. One is an internal effort-based contract, in which the provider supervises the firm internally and the compensation rate depends on the firm’s security efforts once the breach occurs. The other one is an external effort-based contract, in which the external monitoring organization monitors the provider’s and the firm’s efforts.

ABSTRACT. This paper draws on an organization-environment co-evolutionary perspective to conduct a longitudinal single-case study on the digital transformation of Midea Group, aiming to reveal the interaction mechanism between traditional manufacturing enterprises and their micro- and macro-environment. The study finds that similar to the evolution of species, the transformation starts with the perception of internal and external environmental stimuli, which triggers micro adjustments of "cognitive model-organizational elements" at the organizational level, and later triggers macro adjustments of "organizational elements-competitive position" at the environmental level. It becomes the basis for a new round of internal and external environmental stimulation perception. The research model provides ideas and practical guidance for the dynamics, process, and results of the digital transformation of traditional manufacturing enterprises.

ABSTRACT. This study explores the influence of three specific community-based mechanisms on trust and transaction intention from three dimensions of relational governance (relationship norms, conflict resolution and mutual dependence). At the same time, this study compares the differences effect between consumers and prosumers. Taking the second-hand trading platform Xianyu as an example, structural equation model was used to analyze the data based on 721 valid questionnaires. The results show that interest group, feedback mechanism and dispute resolution mechanism all have significant positive effects on seller trust. In addition, the impact of dispute resolution mechanism on trust in seller is higher for prosumers than for consumers. Current research focuses on the impact of social attributes of community on users, this study explores different specific community-based mechanisms. This study extends the previous research on community-based governance, explores the boundary conditions of community-based governance.

ABSTRACT. 物联网、大数据、人工智能等数智技术与制造业深入融合，促进了C2B个性化智能定制的发展。本文引入尚品宅配C2B个性化定制案例，研究数智化技术驱动的C2B个性化智能定制平台运营创新。结果表明：数智化技术驱动O2O产品设计机制创新，可提升前端需求资源连接效率、产品设计数据挖掘效率、产品智能设计决策效率，推动产品数智化设计；数智化技术驱动订单管理机制创新，可提升中端拆单流程连接效率、订单管理数据联通效率、订单智能管理决策效率，推动订单数智化管理；数智化技术驱动数智生产机制创新，可提升后端生产资源连接效率、产品生产数据分析效率、产品智能生产决策效率，推动产品数智化生产；以产品数智化设计、订单数智化管理、产品数智化生产为目标，实施数智化技术驱动的C2B个性化智能定制平台运营创新，可促进产品C2B个性化智能定制。研究成果对推动传统企业数智化转型具有重要指导意义。

ABSTRACT. Internet have radically transformed how people live their lives, especially in the COVID-19 pandemic. People rely on various Internet platforms in daily life, such as e-commerce platforms, social media platforms, etc. With the time going, Internet platforms have accumulated more and more personal information about their users or customers. In order to improve user experience and increase their profits, Internet platforms are willing to open their interfaces to third parties and share users’ information with them. Internet platform gains revenue by providing services for users, the users get utility by Internet platform. At the same times, the third parties enhance users experience by providing diversity services, and gain revenue and users' data. However, such kind of sharing may infringe users’ privacy and then erode users’ trust, especially for those users with higher privacy concerns. This paper tries to find out how Internet platforms can walk a fine line between sharing users’ information and protecting their privacy. Firstly, we analysis the cost-benefit of stakeholders. Then, we propose a pyramid model that captures the interaction between the users, Internet platform, and third parties. Specifically, we focus on the influence of user privacy sensitivity on information sharing behavior of Internet platform. We construct a three-stage dynamic game model with complete information to obtain the threshold value of third parties under the equilibrium state. The result shows that unrestrained sharing of user’ information with third parties will bring negative effects for Internet platforms and finally reduce their profits. The most advantageous way is to develop differentiated privacy policies for users with different privacy preferences. Those users with higher privacy concerns can pay some membership fee to enjoy the service provided by an Internet platform without the disturbance from various third parties.

ABSTRACT. 原生广告旨在与周围内容的融合，可以在不破坏消费者浏览体验的情况下实现广告的目的。但与原生广告相比，看起来与编辑内容不同的展示广告可能会因为高强度的曝光吸引更多的关注并获得更高的品牌认知度。因此，展示广告的引入可能会对移动电商平台原生广告产生影响。本研究以中国某大型网上订餐平台的在线广告准自然实验为研究情境，基于双重差分的方法实证检验平台层面移动电商平台引入展示广告对原生广告效果的影响。

ABSTRACT. Channel selection is a critical trade-off for the information products that are characteristic with network externality. The current work develops the optimization models for the three channel strategies in the two-sided market: direct channel, platform channel and hybrid channel. The mathematical results show that, if the intensity of network externality for the online platform surpasses that for the information products, the hybrid channel strategy domains the other two strategies, otherwise, the direct channel is the optimal choice. The current work provides decision support for the information products firms on channel selection in the context of two-sided market.

ABSTRACT. 数字平台连接供需两端使得交易在线化，提高市场效率的同时也放大了平台产品的质量与服务问题，平台中的用户投诉数量居高不下。投诉会损害供给端的声誉并可能影响其在平台中的曝光率，进而影响其长期发展，最终不利于平台经济健康发展。因此，本文认为用户投诉会对平台供给端产生影响，包括提高服务水平和增加商品发布数量。基于此，本研究以中国某租房平台为研究情境，关注用户投诉行为，使用双重差分法（DID）检验用户投诉对平台供给端所产生的具体影响。实证结果支持如下结论：（1）用户投诉会导致平台供给端提高服务水平，但不会明显增加房源发布数量；（2）房源相对价格越高，受投诉的平台供给端的服务水平提升越有限，新增房源数越少；（3）所处商圈竞争压力越大，受投诉的平台供给端不会明显提高服务水平，但新增房源数越多。

ABSTRACT. Financial fraud has occurred from time to time and disrupts the normal operation of the market and hinders the healthy development of the accounting industry. To address this issue, the US Securities and Exchange Commission (SEC) initiates lawsuits against companies that violate accounting standards and published the records of these lawsuits on Accounting and Auditing Enforcement Releases (AAER). Researchers and practitioner from all kinds of area can then extract different information from it. However, at the moment, most information extraction algorithms depend on large human labor to define rules or mark tags for the training data, which is not feasible for this task. Our study proposes an end-to-end information extraction framework, the context-based one-shot information extraction system (COIES). Our experiment shows that the results of COIES approach human performance which requires only one example with the minimal prior knowledge input.

ABSTRACT. Based on the technology of knowledge atlas, the association network of A-share listed companies is constructed by using multi-source heterogeneous data of listed companies, using the top-down construction method, through ontology modeling, knowledge extraction, knowledge fusion, knowledge storage and other steps. On this basis, we use graph algorithms such as similarity calculation and community discovery to mine potentially related enterprises and identify risk communities. Finally, through key processing such as node merging and relationship synthesis, the correlation graph of listed companies is transformed into a risk contagion graph, and a risk contagion model based on personalized PageRank algorithm is proposed to analyze the risk contagion problems caused by different types of risk sources. From the perspective of Enterprise Association, this paper gives different risk weights to different types of relationships according to their different importance in the process of risk contagion, and uses multi case study methods to analyze the process and results of risk contagion in three scenarios: management fraud, negative changes in equity relations, and negative changes in shareholding and holding relations, in order to make a more comprehensive, detailed, and more effective simulation and prediction of risk contagion, Provide new ideas and technical support for the risk management of financial supervision departments.

ABSTRACT. 本文研究了公众关注对分析师访问的影响。为了建立因果关系，本文使用胡润百富榜作为外生冲击来识别公众关注的变化，使用双重差分法，本文发现公众关注显著提高了分析师对上榜公司的访问数量。这种关系在信息环境复杂的公司、股票收益波动较大的公司以及非国有上市公司中更为明显,表明公众关注对分析师访问量的正向影响是通过投资者对分析师服务的需求这一渠道实现的。进一步分析表明，分析师对公司的访问提高了分析师盈余预测的准确度。本文结果强调了公众关注对企业信息环境的重要性。

ABSTRACT. 本文探讨了全球经济政策不确定性对加密货币收益稳定性的影响。本文的研究结果表明，在样本期内加密货币可以作为抵御经济政策不确定风险的避险资产，经济政策不确定性越高，加密货币的流动性波动和收益波动越低。此外，加密货币的流动性波动会显著的抑制经济政策不确定性对加密货币收益波动的负向影响，但是这种抑制作用在COVID-19大流行期间更为显著。本文还发现了一些基于国家层面经济不确定性以及基于加密货币特征的异质性结果。

ABSTRACT. Constructing a proper SME credit risk assessment index system and developing a predictive model that delivers with excellent performance will reduce the uncertainty of financial institution loans and ensure better development of supply chain finance (SCF). This paper proposes an integrated SME credit risk assessment index system and an innovative prediction model called XGBoost-RF to forecast SMEs’ credit risk. The results indicate that (1) the final assessment system offers 31 specific indicators for 5 dimensions, namely, SME characteristics, core enterprises characteristics, item characteristics, the operational status of the supply chain, and the macro-environment. (2) The assessment system and the XGBoost-RF model proposed in this paper exhibit excellent predictive performance in SCF. (3) Debt-paying ability most affects SMEs’ credit risk. (4) The effects of some new influential factors (e.g., the level of information processing and top management team concurrent posts) on SMEs’ credit risks are also verified.

ABSTRACT. 作为农业金融生产系统的重要组成部分，借贷是促进农村发展的有效手段。鉴于农村紧密的社会关系，家庭选择借贷渠道时会参照同伴行为。考虑信用市场正式与非正式借贷并存的二元特征，研究选取中国家庭追踪调查（CFPS）数据，采用面板Probit模型探讨同伴家庭对农村家庭借贷渠道决策的影响。结果表明：农村家庭借贷渠道选择存在显著正向的同伴效应；但来自互联网、报刊杂志与短信渠道的信息会削弱正式借贷中的同伴效应，来自他人转告的信息则强化了非正式借贷中的同伴效应。异质性分析发现，地位高的家庭更容易受同伴影响选择正式借贷，而对于有商业保险的家庭，非正式借贷中的同伴效应更强。考虑样本选择、反向因果和渠道干扰问题后，研究结论保持稳健。

ABSTRACT. This study employs value-added, relevancy, timeliness, completeness, and interestingness to measure the quality of disclosed information, introduces engagement to the context of crowdfunding, and examines how information disclosure quality will influence investors’ investment decisions in crowdfunding. The analysis results show that crowdfunding investors attach importance to value-added, relevancy, timeliness, and interestingness among the five measurements of information quality. Affective and cognitive engagement are supported to significantly impact investors’ retention, while behavioral engagement is not supported.

ABSTRACT. By the end of 2020, the number of credit cards issued in China had reached 778 million, and with the impact of the new crown pneumonia epidemic, the default rate of credit cards in China has been increasing year by year. For the banking industry, credit card business is an important part of its personal loan business, and it is crucial to forecast the default probability of cardholders. A robust predictive model is not only a useful tool for banks to make judg-ments at the time of credit card application, but will also help customers to realize that this is an action that may harm their credit score, thus achieving the effect of default prevention beforehand.With the continuous enrichment of cardholder information, traditional statistical tools can no longer meet the prediction needs, and machine learn-ing methods suitable for big data analysis have started to be applied in the field of credit card default prediction. Among them, the Gradient Boost Decision Tree (GBDT) algorithm has emerged with various improvements based on GBDT, including XGBoost, LightGBM and CatBoost, because it can achieve better results in different scenarios. Among them, CatBoost is the improved version of GBDT with better comprehensive effect. Based on CatBoost al-gorithm is widely used in e-commerce, disease prediction and other fields, this paper uses CatBoost algorithm to predict credit card default probability.In this paper, the cardholder information of American Express credit card during 2017 is selected as the dataset, the variables are made easy to read by changing the data format, exploratory analysis is performed on 190 variables, various visualizations are tried, the categorical variables among them are selected, and Catboost is used to train the classifier and make predictions, and good accuracy is achieved.

ABSTRACT. 跨组织信息系统（IOIS）产生并支持跨组织协作关系，临近度被视为跨组织关系的重要影响因素，但已有研究尚未对其影响机制达成一致结论。本文基于复杂网络方法进行实证分析，探究IOIS中临近度对跨组织协作关系的影响机制并识别三项调节作用。结果表明：IOIS中，临近度对跨组织协作关系具有“倒U型”的非线性影响；地理距离、结构型社会资本与关系型社会资本显著调节倒U关系并降低最佳临近度水平；两种社会资本均正向影响跨组织协作关系。

ABSTRACT. The COVID-19 has had a severe impact on global health and the economy. Policy interventions conducted by the government play an essential role in curbing the spread of the COVID-19. In this study, we propose a policy support system based on deep reinforcement learning (DRL), to provide scenarios of artificial intelligent practices for crisis management. The preliminary results show that the proposed system outperforms the benchmark metrics and policy implementations in the real world significantly, regarding the infection rate and normalized discounted cumulative economic gain from a long-term view.

ABSTRACT. 新能源政策是我国国家长期战略政策，具有随时间变迁、地域差异和多主体协同制定三个特征，对这三个方面开展研究有助于深入认识我国新能源政策制定的规律和实现政策智能。本文基于2011年1月至2022年1月期间，由中央、各部委以及各省区颁发的共计1937份新能源政策文本，采用文本挖掘和共现网络分析方法，从政策的发布时间、发布地域和发文主体三个维度对新能源政策进行分析，揭示我国新能源政策的时空演化规律和发文主体协同特征。

ABSTRACT. Digital government is a new governance model that emerged in the era of advanced information technology. Providing public services is one of the government’s foundational capacities. This research aims to figure out key factors that may contribute to the varying levels of public service capacity of provincial governments in China. With the technology-organization-environment (TOE) framework, this research identifies five key influencing factors for the public service capacity of digital government. By using the qualitative comparative analysis method, this research figures out three solutions that can result in a high level of digital government public service capacity. The results suggest three typical development paths of digital government, including technology-economy driven provinces, platform-organization driven provinces, and organization-demand driven provinces. This research has implications for helping provincial governments to improve and allocate resources to advance their digital government public service capacities.

ABSTRACT. 第四次工业革命带来了前所未有的城市治理变革，我国各超大城市都在全面探索数字化转型的这一技术治理导向的城市治理创新之路。与此同时，城市管理体制发展正在经历“重心下移”的趋势。在此双重逻辑下，行政层级上如何回应提升政府治理能力的需求？从技术-组织耦合视角，通过对上海市城市运行管理中心“一网统管”实践案例的深入分析，系统考察了数字政府平台的区－街统合中需要创新的四大清单，职责清单、任务清单、流程清单和人员清单，及其相应的耦合悬浮四大实践困境表征，即责任分担困境、上下整合困境、部门协同困境和事务处理困境。并据此提出相应的优化路径，以期有效解决在数字化转型背景下的城市治理困境。

ABSTRACT. This paper adopts the idea of policy informatics text analysis, uses the lexicon-based method of sentiment analysis to mark the sentiment of citizens' calls in Beijing 12345 citizen hotline, and combines the theoretical framework of "expectation dis-confirmation" to measure the public satisfaction. The main findings are as follows: when the public is interfered with by public events, the public's demand for government services increases, and this service supply gap will lead to a decline in the public's satisfaction with government services; The dispatching strategy and sentiment comfort of call operators also play an important role in improving public satisfaction.

ABSTRACT. 围绕安徽省政府网站集约化平台统一知识问答库体系应用场景建设，本文阐述了统一知识问答库的基本概念和特征内涵。在面向社会公众政务咨询领域的各类政府网站、政务服务、12345热线、政务新媒体等平台互动渠道知识问答库集群构成和内容分析基础上，从知识问答库集群集约共享和有序化组织管理服务的角度，构建了基于核心元数据集的统一知识问答库体系框架，并在参考借鉴已有政策文件和元数据标准规范的基础上，建立了统一知识问答库核心元数据集及其对应的知识库辅库体系。

ABSTRACT. 为缩小我国城乡“数字鸿沟”，在农村公共服务供给呈瓶颈之态时，数字赋能成为推动农村公共服务高质量供给的方向所在，但中国国情下数字化作用于农村公共服务供给逻辑尚不清晰。本文基于数字赋能视角，运用案例分析法探讨了其与农村公共服务供给之间的因果关系。对陶坝村“为村”平台的实践历程中发现：参与路径闭锁、政绩显示过度、单中心治理格局与信息化水平滞后是服务质量差距产生的根源；数字具有的交互性、共享性、整合性、敏捷性与去中介性对供给环节中表达、决策、匹配、评价进行赋能而弥合质量差距。研究有助于从我国农村实际情况出发阐明数字化与农村公共服务质量的内在联系。

ABSTRACT. This paper explored the health information needs of the elderly from two perspectives: information demanders and information providers. It crawled data from the questions and health education sections on the online health communities. LDA model was first applied to cluster the unlabeled questions set and obtain five themes. Then the dataset of article titles with five categories was used to train the classifiers which were CNN-BiLSTM and FastText. After comparing the matching degree between clustering themes and classification results, this paper put forward some suggestions for elderly caregivers and information providers to assist the elderly health management.

ABSTRACT. 基于信息系统有效使用理论，本文采用混合研究方法结合多源数据，通过三个子研究设计揭示了消费者参与的形式、测度体系与影响效应。研究表明，移动医疗服务价值共创的消费者参与是技术赋能下消费者与各利益相关者资源交互与整合的过程，包括消费者-提供者参与、消费者-消费者参与、消费者-技术参与，这三个二阶维度构成了消费者参与的层次结构。同时，基于消费者参与形式构建了消费者参与的测度体系，并利用专家与用户评价、信效度检验与外部效应检验验证了测度体系的有效性与稳健性，表明本文所构建的测度体系可以为后续实证研究提供可操作化的测量量表，也为丰富健康情境下的价值共创研究以及信息系统有效使用理论作出了相应贡献。

ABSTRACT. 缺血性脑卒中作为一种常见的急性脑血管疾病，是导致成人死亡和致残的重要原因。本研究旨在利用基于多源数据融合的机器学习算法，预测缺血性脑卒中患者的临床药物治疗风险。通过融合国际脑卒中试验数据集，对比随机森林、Logistic回归和梯度提升决策树算法的预测效果。其中，梯度提升决策树的召回率达到91.6%，AUC为0.832，效果最佳。实验结果表明，基于多源数据融合的机器学习算法在缺血性脑卒中药物治疗风险预测中具有较好的适用性。

ABSTRACT. Medical crowdfunding has been developing in many countries and helping people cope with the challenges of medical issues. This study explores the donation intention of medical crowdfunding users on online platforms. Based on dual-system theory, empathy-helping hypothesis and trust building model, a model was constructed to explore the factors affecting the donation on medical crowdfunding. We used a survey to collect data from people who have donated to medical crowdfunding projects. The analysis results reveal the influence of negative emotion appeal and familiarity on empathy, and the influence of structural assurance and third party seal on perceived credibility. This study not only expands the theoretical research on users’ behaviors in the context of medical crowdfunding, but also provides enlightenment for optimizing the process of medical crowdfunding projects.

ABSTRACT. This paper analyzes the user's comments on an mobile platform of medical and senior care (MPMSC) to understand the user's concerns and help the platform operators improve the platform functions and services. Based on LDA (Latent Di-richlet Allocation) model and content analysis method, this paper analyzed 2569 user reviews about MPMSCs in Apple App store, and a theme framework of user review on the MPMSC is developed. The framework includes four first-level topics such as overall recognition, service convenience, service quality, and system quality, as well as 11 second-level topics. The conclusion enriches the relevant theoretical research of mobile application evaluation, and provides man-agement enlightenment for the platform and government to promote the development of MPMSC.

ABSTRACT. 为促进智慧医疗在全国范围内的推广，探究智慧医疗发展进程中的医患价值共创问题，揭示智慧医疗在长期范围内的演化进程。本研究从价值共创视角切入，以演化博弈理论为基础，引入风险系数和意愿系数，基于博弈论模型研究智慧医疗推广中医疗机构和用户的关系，并运用Matlab软件对博弈双方的决策行为进行模拟验证。研究表明，降低医疗机构和用户参与智慧医疗价值共创的风险系数，并增大意愿系数有利于推动价值共创视域下智慧医疗的推广

ABSTRACT. Recently, short-video platforms have become popular among the middle-aged and elderly. While some studies proposed that excessive use of short-video platforms may lead to many psychological problems, little knowledge is about how to help the middle-aged and elderly use platforms reasonably and gain subjective well-being. Drawing on generativity theory and social cognitive theory, this study demonstrated that middle-aged and elderly users who have a higher level of need to be needed would achieve a higher level of subjective well-being and build more social interaction on short-video platforms. Especially, through building more social interaction on short-video platforms they can achieve a higher level of online subjective well-being. Moreover, when middle-aged and elderly users received more family cognitive support, the positive relationships of need to be needed with subjective well-being and social interaction on short-video platforms would be stronger, however, the positive relationship between social interaction and subjective well-being would be weaker.

ABSTRACT. 本文使用信息融合技术和熵权-TOPSIS方法，综合利用了在线医疗平台上的患者咨询文本信息和医生患者交流记录两类数据，从专业匹配和医生服务质量两方面出发，构建了一个为在线医疗平台上的患者进行医生推荐的模型。利用好大夫平台上的相关数据，验证了信息融合的模型得到的推荐结果要优于基础模型，同时给出了推荐的医生服务质量评价得分

ABSTRACT. Artificial Intelligence (AI) can be used in marketing services and dramatically transforms consumer experience. However, little is known about the differences in interacting experiences between AI and human service and how interacting experiences impact consumer trust toward the AI/human. Based on cognitive appraisal theory, this study investigated consumer experience and brain responses underpinning passive interaction with AI service versus human service when participants faced objective or subjective tasks in e-commerce using a behavioral experiment and Event-Related Potential (ERP). The behavioral results showed that AI service experience evoked a more negative consumer emotion and led to a lower trust attitude towards AI service solutions than human service, moderated by task type. The ERP results showed that interacting with AI service evoked larger amplitudes at P2 and LPP than with human service, which suggested that consumers automatically put more attention on AI service at the subconscious stage and purposively allocated cognitive resources to regulate the negative emotion elicited by AI service at the conscious stage. Consumers' trust towards solutions of AI/human service and the moderating effect of task on trust in ERP experiment were consistent with the behavioral experiment. Moreover, the moderating of objective or subjective tasks on emotional experience appraisal was only found in the late emotional regulation stage (reflected by LPP). The findings offer important theoretical and practical implications for implementing the AI service in e-commerce and suggest that reducing the bias against AI is very important for AI agent providers and e-retailers.

ABSTRACT. 对消费者购买行为及其内在机理展开深入研究，能帮助企业及营销人员更有效的了解消费者需求，实施营销计划，完成企业目标，实现企业愿景。目前关于在线视频广告的广告效果研究，许多学者将目光聚焦于广告和视频的相关度，包括广告产品和视频主题的相关度，广告类型和视频类型的相关度，广告音乐和视频音乐的相关度等。研究发现，广告相关度对消费者的品牌态度和购买意愿会产生积极影响。但大多数都是采用传统测度的行为学实验进行研究的，也就意味着它无法揭示消费者在观看不同相关度的广告时，大脑对广告信息的认知加工过程。本文探讨在线视频广告在内容方面的相关度对消费者品牌态度的影响，以及他们如何受到情绪、加工流畅性和说服知识的影响。通过结合使用功能性磁共振成像技术和问卷调研法，本文得出以下结论： （1）广告-视频的内容相关性会影响消费者对广告中出现商品的品牌态度。相关性程度越高，品牌态度越好； （2）广告-视频的内容相关性会影响消费者对广告的情绪，进而影响对广告中出现商品的品牌态度。相关性程度越高，情绪越积极，品牌态度越好； （3）广告-视频的内容相关性会影响消费者对广告信息的加工程度，进而影响对广告中出现商品的品牌态度。相关性程度越高，信息加工流畅性越好，品牌态度越好； （4）广告-视频的内容相关性会影响消费者大脑杏仁核和纹状体ALFF的激活。相关性程度越高，激活越强烈。 （5）加工流畅性和杏仁核激活情况存在显著正相关关系，加工流畅性越高，杏仁核激活强度越大。 （6）随着视频播放，消费者进入视频情境，杏仁核ALFF激活程度逐步降低。 本研究在理论层面，不仅从管理学角度解释视频-广告的内容相关度为什么会影响消费者的品牌态度和购买意愿；而且从神经角度探寻广告相关度对消费者购买意愿影响背后的神经机制，丰富国内学者对在线视频广告效果的研究。最后，在现实层面，可以给企业的广告的内容制作和投放策略提供理论支撑和建议。

ABSTRACT. 虚拟现实技术（Virtual Reality，VR）的发展和应用场景的拓展正在改变人们的日常生活。虚拟现实+教育、虚拟现实+医疗、虚拟现实+地产等领域的成功应用，尤其在元宇宙（Meta）的概念提出后，使针对VR及其在各领域的应用研究探索越来越多。本文从消费者在虚拟现实环境下的感官体验和消费决策这一问题出发，对虚拟现实下的感官体验、消费决策和相关的神经科学研究方法的相关文献进行了整理和分析，提出了以“感官超载”为切入点的未来研究方向建议。

ABSTRACT. We present the COCOATREE (Corpus for Cognitive Analysis of Text Reading with Eye-tracking and Electroencephalography), a dataset combining eye-tracking and EEG (electroencephalography) recordings while reading financial text. To construct the dataset, we recruited 11 native Chinese speakers fluent in English to read our text materials in an experimental environment. We recorded their gaze and brain activity data for two types of reading tasks (i.e., normal reading and task-specific reading) and three types of financial text (i.e., English annual report, English financial news, and Chinese financial news). To our best knowledge, this is the first dataset of simultaneous eye-tracking and EEG recordings for analyzing the reading behaviors of native Chinese speakers in finance and economics. Our dataset also provides a valuable resource for machine learning of cognitive natural language processing models.

ABSTRACT. 本文采用事件相关电位(ERP)方法，探讨了商品的价值和需求特征对选择过载效应的调节效应及神经机制。主观报告显示，过多的选项数目提高了被试的商品满意度，但也提高了感知选择难度和选择放弃意愿，存在显著的选择过载效应。脑电结果表明，实用型商品的选择过载效应更强，价值和需求的调节脑机制存在差异：较享乐型商品而言，实用型商品的选择过载源于注意力消耗过载；而较低价值商品而言，高价值商品的选择过载源于判断力消耗过载。

ABSTRACT. Prior research has indicated that people with similar personal traits (i.e., personalities, interests, taste) tend to select the same type of profile pictures and profile pictures may influence consumer’s perception and behavior. The current short paper offers a novel insight into the match effect of profile pictures, that is, when the types of profile pictures match, buyers are more likely to purchase the products or services provided by sellers. Based on research stream on profile pictures and the theory of homophily, we propose that the match effect of profile pictures leads to increased perceived homophily and thus drives consumers’ purchase behavior. An initial study has been conducted to confirm the match effect of profile pictures. Two experimental studies, one of which will adopt neuroscience method – event-related potentials (ERPs), are proposed to provide more stringent evidence and examine the possible moderators (i.e., information type and information quality).

ABSTRACT. According to the Elaboration Likelihood Model (ELM), there are two routes when individuals process information, the central and the peripheral route, and involvement is an important factor affecting the choice of routes. However, the influence of health involvement on the choice of patient information processing route on online health consultation platform is still elusive. Here, we first used behavioral experiment to identify the impact of health involvement on patients' information processing routes, and then conducted a functional magnetic resonance imaging (fMRI) experiment to measure the activation of ventromedial prefrontal cortex (VMPFC) and ventral striatum (VS), the two key brain regions that reflect individual cognitive resource allocation. The results showed that the VMPFC and VS brain regions were significantly activated when health involvement was low, suggesting that patients chose the central route. When health involvement was high, no relevant brain regions were activated, indicating that patients chose the peripheral route.

ABSTRACT. After service failure, the occurrence of chatbots often brings bad service experiences to people. In order to improve consumer satisfaction when interacting with chatbots in service recovery, we design an online experiment to explore appropriate communication style designs for chatbots, also considering the moderating effects of consumers’ relationship orientation and task complexity. Results show that during service recovery, consumers are more satisfied with social-oriented chatbots rather than task-oriented ones. Besides, we explore the mechanism by which the communication style affects consumer satisfaction in service recovery by introducing cognition-based trust and affect-based trust as mediating variables.

ABSTRACT. The rapid growth of virtual reality (VR) technologies and computing power has given birth to immersive platforms, where users can immerse in a 3D virtual world and engage in various activities such as job training, remote collaboration, and innovative marketing practices. Through the lens of self-perception theory, this paper investigates how avatar design, i.e., user-avatar similarity, affects users’ self-concepts and shapes their behaviors in immersive VR. The preliminary experiment reveals that higher avatar-user similarity leads to higher task engagement in general. Furthermore, while a similar avatar promotes users to regulate their behaviors and achieve better performance in a procedural task, high similarity also inhibits users’ creativity by invoking habitual thinking, resulting in worse performance in generating original ideas in a creative task. This study is expected to contribute to HCI literature by revealing the value of avatar design and providing new perspectives in improving users’ experience in the immersive virtual world.

ABSTRACT. 问答社区内促进问题浏览向回答的流量转化至关重要。以知乎的2085条问题为样本，应用模糊集定性比较分析，从注意力分配视角对问答社区的流量转化进行组态研究。研究发现：高流量转化率问题的组态路径包括低竞争-强从众刺激与低竞争-高中心度刺激的简洁型问题，与非高流量转化率问题的组态路径并非完全因果对称；知识结构化程度不同的问题实现流量转化的组态存在差异。结论有助于促进平台流量转化，优化问答界面设计。

ABSTRACT. Based on the relevant literature on the elaboration likelihood model and vicarious learning theory, this study investigates the impacts of the danmaku sent by co-viewers on consumers’ purchase intention in E-commerce live streaming. An online experiment was designed and conducted to examine the effects of danmku on consumer behavior in a simulated E-commerce live streaming setting. It is found that the accuracy and consistency of danmaku sent by co-viewers increases consumers’ purchase intention by improving consumers’ vicarious learning. Meanwhile, the credibility of danmaku affects consumers’ purchase intention via eliciting consumers’ resonance. Also, the findings show that age has a moderating effect on the impacts of vicarious learning and resonance on consumers’ purchase intention.

ABSTRACT. In live-streaming e-commerce shows, the reputation of influencers plays a central role because it affects the brand-side price negotiation and the consumer-side long-run profitability simultaneously. Building on the career concern literature, this paper examines the effect of the influencer's reputation on price negotiation. Our analytical and numerical results suggest that the price obtained from the negotiation is affected by the reputation concern non-monotonically. An influencer with a greater reputation concern may get a smaller discount. Besides, the brand, the influencer, and consumers are better off simultaneously if the influencer has a moderate career concern.

ABSTRACT. 疫情催化下，企业社交媒体技术广泛应用（ESM）为员工沟通和知识共享提供了新场景。基于心理契约理论并利用我国358名钉钉用户数据进行实证分析，我们发现，社区帮助动机与员工显性和隐性知识共享行为均积极相关，但组织帮助动机仅与隐性知识共享行为负相关；组织依恋和社区归属感与两种知识共享行为均显著正相关。进一步的中介分析表明，社区归属感能够增强社区帮助动机进而间接促进显性和隐性知识共享。通过纳入“个人-集体”心理关系视角，本研究突破了知识共享的个人“成本-收益”分析框架，能为企业制定ESM知识管理策略提供新思路。

ABSTRACT. 本文结合TOE框架，采用模糊集定性比较分析方法，从组态视角揭示提升企业数字化转型的关键路径。主要研究发现是：单一因素无法构成高水平数字化转型的必要条件，高水平企业数字化转型是多重技术、组织、环境因素共同作用的结果；具体地，四条不同的组态路径能实现高水平数字化转型，它们均是技术、组织、环境三方面因素的不同组合，而且，信息技术基础设施和CEO教育是关键的核心要素和辅助要素，行业竞争和研发投入之间存在匹配效应。

ABSTRACT. Strengthening the construction and management of government websites, improving the government's online duty performance ability and service level, and building an overall linkage and efficient online government benefiting the people are the key links to improve the efficiency of government services. With public increasing demands for search function, government websites are expected to improve dramatically search and services to provide what the public want from government websites, including needed information, working procedures, and FAQs, that is “what you search is what you want.” Based on the practice of the website of the People’s Government of Beijing Municipality, this research aims to investigate strategies for improving search and services of government websites. From the perspective of both government functions and public appeal, our website designed the common thesaurus used in daily life through technology innovation, and constructed the intelligent scenario with a high degree of match between the government service to provide and the demand keyword of public search. Again, creating the tagging manage system of the website content, our website implemented the search engine based new service concept incorporating social media.

ABSTRACT. 本文基于社会技术系统理论探究知识付费情境下消费者价值共创行为的影响因素及其对消费者满意度和持续付费意愿的作用，并结合模糊集定性比较分析法探究因素组合对消费者价值共创的影响。结果表明：社会因素（认同感、互动性、知识创新性与稀缺性）与技术因素（元发声与个性化）均能够促进消费者价值共创，进而提升消费者满意度和持续付费意愿，同时存在四类触发消费者价值共创行为的构型模式。

ABSTRACT. 电子商务推荐系统是一个非常重要的工具，它可以帮助用户过滤无用信息并且找到他们感兴趣的资源(如电影、音乐、书籍等)。本文提出一种基于协同排序的电子商务个性化推荐方法，首先引入矩阵分解技术对评分矩阵进行分解，同时获取给定用户和商品的特征向量，形成用户-商品对的特征表示。然后应用排序学习算法LambdaRank来学习排序模型，并与MF和RankWLT两个算法的实验结果进行比较，实验证明了本算法的有效性，能有效提高Top-K推荐性能。

ABSTRACT. 针对多指标组合赋权问题，提出了基于排序学习的主客观组合赋权方法，目的是使指标赋权的结果同时兼顾主观权重和客观权重。该方法首先构建指标和评价对象的初始偏序关系；然后构建指标和评价对象的预测偏序概率以及排序损失函数；最后以预测概率与真实概率的差距最小为优化目标，建立综合决策模型。同时，对8家快递型物流上市企业2021年度财务数据进行了实证分析，通过实例说明了该方法的合理性和有效性。

ABSTRACT. 在线信誉反馈系统是电商平台的重要服务手段，而买家评价的可信度则是在线反馈质量的关键问题。本文基于自然语言分析技术，提出了一种个性化的在线信誉反馈机制，构建买家信誉评估模型Ces-TA。通过文本语义分析，挖掘基于商品特征的个性化评价指标，使得用户评价更具针对性。通过用户评价的文本情感分析，评估买家评价的一致。基于大数据分析评估买家的价格偏好和情感偏好，借此修正买家信誉，从而使买家信誉更具客观性，最后的实验展示了算法效果。

ABSTRACT. 电商直播间是一个多信息源交互的复杂信息环境，弹幕评论是其中重要的信息源。本研究基于ELM理论构建了电商直播中弹幕信息质量对消费者购买意愿的影响路径模型。借助结构方程模型，通过643位电商直播用户的调查数据验证研究假设。结果表明：中心路径的弹幕内在信息质量和边缘路径的弹幕表征信息质量，能够通过影响消费者的感知功利和感知归属进而影响其购买意愿。消费者的购物导向影响着消费者感知价值形成过程中对不同路径信息质量处理的详尽可能性。消费者购物导向和主播知名度在消费者感知功利和感知归属对购买意愿的影响关系中起调节作用。本研究丰富了电商直播和弹幕相关研究的理论成果，延伸了ELM的应用范围，发展了ELM的理论内容，发现了不同消费者个体在不同主播直播间中受弹幕影响下形成购买决策的理论边界。研究成果有利于推动平台、商家和主播营销策略的改善，推动电商直播的繁荣发展。

ABSTRACT. This paper examines the impact of information technology investment on the high quality development of enterprises, taking the A-share listed companies in China from 2010 to 2020 as the research object. Further mechanism tests show that corporate innovation and human capital play a mediating role in the relationship between information technology investment and high-quality corporate development, i.e., corporate information technology investment promotes corporate innovation and optimizes human capital structure, which in turn promotes high-quality corporate development.

ABSTRACT. 从宏观经济、传统金融行业、互联网行业和互联网金融行业等维度构建混频数据集，用于测度和预测互联网金融风险水平。研究表明：随着互联网金融从“缺门槛、缺规则、缺监管”过渡到“高门槛、严规则、强监管”，行业风险水平整体呈下降趋势；银行不良贷款、网络借贷、互联网金融指数波动和互联网金融规模对风险变化有较高解释能力；中国互联网金融存在高低风险区制转换，未来会经历温和上升过程，但整体仍处于低风险区制。

ABSTRACT. 文本复杂度是关于企业年报信息质量的一个重要研究方向，围绕这一主题产生了大量研究。 当前的年报复杂度指标混淆了过多信息和冗余信息，并且无法区分二者。鉴于此，本文以计算语言学为基础构建文本压缩率来衡量上市公司年报的复杂度，进而对过多信息和冗余信息进行区分。本文研究了年报复杂度与盈余持续性之间的关系。研究发现。(1) 年报更复杂的公司有更不持久的正收益或更持久的副收益；(2)导致年报复杂的原因主要为冗余重复的信息。

ABSTRACT. 在当前动荡的市场环境下，风险预见性研究于企业而言至关重要。为明晰国内外在弱信号领域的研究现状，并解决企业风险管理领域弱信号运用过于泛化的问题，本文对弱信号相关文献进行梳理与分析的同时，又将弱信号引入企业舆情风险预警领域。首先，运用文献计量法与CiteSpace软件对2000年至2022年弱信号的研究趋势和关键词共现情况进行分析。其次，基于弱信号和企业舆情风险的特性，定义企业舆情风险预警领域的弱信号。最后，结合弱信号关键技术从弱信号搜集、弱信号分析、弱信号传递及弱信号利用四方面探讨弱信号在企业舆情风险预警过程中的应用。基于以上，提出未来可研究的方向，为后续相关研究和企业实际应用提供一种新的视角和内容参考。

ABSTRACT. 为解决港口供应链企业核心信息泄漏问题，降低企业信息安全风险，借助反竞争情报体系与大数据技术构建港口供应链企业反竞争情报模型，为企业提供完备的信息保护机制，对实现港口供应链企业创新发展与转型升级具有重要意义。本文对港口供应链企业信息安全风险进行界定，引入“风险流”、“情报流”，提出“五流一体”港口供应链业务结构与信息共享平台模型框架，实现港口供应链企业间“信息流-风险流-情报流”的交叉互动，构建基于信息安全风险的港口供应链企业反竞争情报机制模型。

ABSTRACT. 为了衡量供应商能否持续改进技术水平，保障供应链的安全运作，本研究基于多种来源的数据对企业的技术创新能力进行评估，构建了分类评估框架和技术创新风险评价指标体系。为验证所构建指标体系的有效性，本研究选取某集团的143家不同规模和类型的供应商为样本，使用层次聚类法、熵权赋值法、文本挖掘方法等研究方法，通过构建包含技术能力、人力资本与研发条件3个核心维度，19个关键指标的指标体系，评估供应商的创新能力。针对其中68家重要供应商，模型测算结果与专家打分相关性为0.58，总体一致性较高，在评估上千家供应商的技术创新风险时有一定优势。

ABSTRACT. 本文利用2012年至2016年间上市金融企业与投资者的网络互动数据，考察投资者关注度对股票收益率的影响以及由此带来的投资风险。本文分别以投资者对上市公司的关注次数、关注的细度以及情绪的积极程度作为代理变量来度量投资者关注。实证结果表明，对于证券、保险、信托、股份制与城商行四个金融子行业，这三类代理变量均为上市公司股票收益率的风险因子。但对于国有银行，这些现象并不显著。上市公司应充分利用风险因子产生的路径进行干预，借助专业信息平台，积极、全面地与市场沟通，加强自身业务的宣传和解读，提高信息透明性和市场有效性，降低股票波动风险。

ABSTRACT. The survival and development of listed enterprises are influenced by financing, and the financing risk assessment of listed enterprises is negatively affected by uncertainties of evaluation information caused by the difference of evaluation experts' experience. To address it, a financing risk assessment method of listed enterprises based on D-number theory improved analytic Hierarchy Process (D-AHP) and approximate ideal solution ranking method (TOPSIS) is proposed. The evaluation index system and hierarchical structure model are established according to the three financing methods of bond financing, equity financing and hybrid financing of listed financing risk. Then the D-AHP is used to solve the impact weight of each index, and the TOPSIS is used to calculate the expert weight. Finally, the financing risk value of listed enterprises is obtained, so as to give reasonable evaluation results about financing risk impact of listed enterprises.

ABSTRACT. 面向客观了解、掌握创新实体的技术创新能力的客观需求，针对现有研究存在的采用的专利指标有待扩展、未对专利类型进行细分等不足，本文构建了一套基于专利数据的机构技术创新能力测度方法，并以北交所首批上市的16家专精特新“小巨人”企业为例进行了应用研究。本文构建的测度模型涉及机构的外在显性表现（技术创新活跃度）和内在隐性价值（技术创新价值度）两个角度，能够较为全面的反映创新机构的技术创新能力，拓展了现有研究中单纯依靠专利基本计量统计指标开展技术创新能力评估的方法；同时在机构的专利隐性价值方面，本研究针对不同的专利类型，有针对性的设计了专利价值评估指标体系，并通过专家研讨的方式确定数据规范化策略和指标权重，对于未来相关工作的开展提供了数据基础和思想指导。

ABSTRACT. The textual risk disclosures in financial reports, which discuss the company’s potential risks from a forward-looking perspective, were rarely considered in financial distress prediction. Thus, this study explores whether the textual risk disclosure of financial reports can help predict the financial distress or not. To comprehensively extract information from the massive unstructured textual risk disclosures, the textual attributes are utilized to capture the linguistic styles of risk disclosures and an unsupervised topic model is adopted to identify the textual topics (the content of the text) in risk disclosures. Based on the textual risk disclosures of 4,039 financial reports for U.S. energy companies from 2006 to 2020, the empirical results demonstrate that the textual attributes and textual topics extracted from risk disclosures in financial reports can significantly improve the financial distress prediction performance compared with commonly used numerical variables (financial and market variables). Moreover, the textual topics in risk disclosures can provide more information than commonly used textual attributes for financial distress prediction. Last but most importantly, the textual risk disclosures become more and more useful relative to numerical variables as the predicting time horizon becomes longer. This study can help investors and regulators understand how to analyze the textual risk disclosures in financial reports and incorporate the textual information into financial distress prediction.

ABSTRACT. 现有研究强调CIO着承担数字化创新的主要职责，但鲜有探讨CIO如何克服资源困境进而推动组织的数字化创新。本文整合拼创和议题营销理论，采用组态理论的观点，认为在CIO个体特征和组织因素等条件下，CIO可以通过拼创或议题营销成功地实现数字化创新。借助模糊集定性比较分析法(fsQCA)对218对CIO/业务高管匹配样本的问卷数据进行实证分析，结果显示，在不同的条件下CIO通过拼创或议题营销使数字化创新达到较高水平。本文也讨论了关于CIO通过拼创或议题营销实现数字化创新的理论贡献和实践意义。

ABSTRACT. 碳交易平台是碳交易市场的重要部分，基于该平台，交易双方通过碳交易达成优化环境资源配置等目的。站在碳交易平台的角度，如何运用参与碳交易实体的信息为决策双方提供有效情报是平台数字化、智能化的发展方向。本文运用动态博弈理论，考虑不同交易情景，研究贴现因子、边际收益对碳交易的影响，为企业推荐交易阶段、贴现因子及边际收益的最优情报。结果表明：推荐买卖双方在第一阶段（情景一）成交，特别是边际收益越大越可能稳定于情景一；并不是双方贴现因子越大越有利，将其维系在此情景一范围内有利于双方交易获利；最优边际收益底线取决于卖方第一次出价和买方第一次出价。

ABSTRACT. 已有研究表明，信息技术（IT）双元能力能够显著增强组织敏捷性和企业绩效。然而，当前尚未有研究探索如何培育企业IT双元能力。本文从首席信息官（CIO）和高层管理团队（TMT）知识交互角度，研究CIO与TMT知识交互机制（包括结构化和社会化交互）对IT双元能力的影响，并考虑环境动态性的调节效应。本文构建了结构性交流系统、社会性交流系统、环境动态性与企业IT双元能力之间关系的研究模型，利用结构方程模型对347家中国企业的调查问卷进行数据分析和模型拟合。研究结果表明，结构性交流系统和社会性交流系统都能正向影响企业IT双元能力，且两种交互机制在增强IT双元能力过程中存在替代效应；环境动态性正向调节社会性交流系统与IT双元能力的关系，而对结构性交流系统与IT双元能力的关系无调节效应。除此之外，三向交互结果表明，在动态环境下，结构性交流系统和社会性交流系统之间的替代效应会减弱。以上结论拓展了当前对IT双元能力培育的研究，并为企业如何构建IT双元能力提供实践指导。

ABSTRACT. 保护环境、节能减排已成为世界共识。2020年，中国提出双碳政策，力争2030年前实现碳达峰、2060年实现碳中和。本文采取fsQCA的方法，基于我国30个省域2019年的数据，探究数字经济驱动下我国低碳发展路径，分人口、经济社会、资源三个层面考察了人口产业结构、数字经济发展水平、城镇化水平、碳减排政策规制和能源消费结构对地区碳排放强度的组合效应，得到影响地区碳排放强度的两种路径，分别为“数字经济-城镇化水平-低能源消费结构”路径和“数字经济-人口产业结构-碳减排政策”路径。

ABSTRACT. 自2019年生态环境部和发改委联合印发《长江保护修复攻坚战行动计划》以来，长江流域省份已经广泛开展了城市管道修复PPP项目试点工作，以实现污水处理的提质增效。本文从安徽省芜湖市城市管网修复项目管理实践出发，针对现有项目管理模式中存在的修复方案的设计进度与管网修复的施工进度不匹配导致工期长和重复作业的问题，提出基于改进支持向量机的城市排水管网修复方案选择模型，通过收集项目数据，分析验证了模型的有效性和可行性。

ABSTRACT. Due to the large amount of inaccurate information online, the public can be particularly vulnerable to misinformation. This study aims to fuse Knowledge Graph and Artificial Neural Network (KGANN) to detect online health misinformation. Our model transforms a public medical knowledge graph into a neural network so that the knowledge graph can be trained by backpropagation algorithm. The model integrates medical knowledge and feature vectors, and determines their weights during training. We performed KGANN on two public disease-specific datasets (including factual and error information related to diabetes and cancer) to validate its effectiveness. The experimental results show that KGANN significantly outperforms two competing methods in terms of F1 Score and Accuracy, and the operation process of the model can be explained to a certain extent through triples in the knowledge graph.

ABSTRACT. Artificial intelligence system is more and more widely used in enterprises. The research on the application effect of artificial intelligence system is of great significance. Taking safe driving as a scenario, this paper investigates the influencing factors of effective use (improving drivers' safe driving behavior) of intelligent safe driving assistance system in enterprises. Based on the theory of person-environment fit, this paper proposes a research model on how individual factors (i.e., age) and organizational factors (i.e., safety assessment system and workload) affect the effectiveness of the system. Further, we analyzed and verified the research model with the driving behavior data from a large petrochemical enterprise for 4 years. The results show that the intelligent safe driving assistance system is more effective for older drivers. When the enterprise is equipped with the corresponding safety assessment system or the workload is low, the safe driving assistance system will be more effective. At the same time, there are also significant interactions among the above factors. When the workload is large, the safety assessment system plays a greater role. The positive effect of safety assessment system on the effectiveness of intelligent monitoring system is more obvious among older drivers. The negative effect of workload on the effectiveness of intelligent monitoring system is more obvious among young drivers.

ABSTRACT. 全社会用电量可以准确反映国民经济的运行情况，预测全社会用电量能够辅助预测经济发展趋势，为制定宏观调控政策提供参考。如今，高频海量的多源数据为全社会用电量的预测提供了新的思路。本文利用高频气温数据构建了考虑季节和用电行为惯性的“季节-累积气温指数”，并提出了基于多源大数据的混频预测模型（MIDAS-MT-DT）。实验验证了MIDAS-MT-DT模型具有更高的预测精度，且季节-累积气温指数具有提高预测精度的能力。

ABSTRACT. Air pollution is a major issue related to the national economy and the people's livelihood. At present, the research on air pollution emissions mostly focuses on analysis in a specific industry or regions. Enterprise is the most important entity unit of air pollution source. While limited by the amount and granularity of data, there are a few studies on air pollution of industrial enterprises. Especially, there is a lack of research on anomaly of enterprises, the relationships between districts or counties, and between pollutants emitted by enterprises and industries of enterprises. For this purpose, driven by big data of air pollution emissions of enterprises in Beijing-Tianjin-Hebei, the data mining of enterprises’ pollution emissions is carried out, including the outlier detection based on clustering, association rule mining based on Apriori and association degree discovery based on grey association analysis. The results show that: (1) In Beijing-Tianjin-Hebei, industrial enterprises can be divided into six clusters, of which three categories belongs to outliers, which have excessive emissions of total VOCs, PM and NH3 respectively; (2) From the perspective of the association between different data, these districts nearby Hengshui and Shijiazhuang city in Hebei province form strong association rules. (3)From the perspective of the association between different features, the industries affecting NOx and SO2 mainly are electric power, heat production and supply industry, metal smelting and processing industries.

ABSTRACT. 本文首先从CNSR1、CHARLS、CLHLS数据集中筛选与老年慢性病相关的风险指标，利用XGBOOST进行特征重要性排序和SHAP可解释性分析初步构建预测指标体系。然后结合粗糙集理论和信息熵算法进行属性简约。采用集成学习的模型进行预后预测，针对离线模型和流数据不匹配的问题，引入KL散度作为衡量新数据和源训练数据的差异度，并通过同构迁移学习适时改进模型。为提高老人晚年的生活质量，慢性病的康复过程需要家庭、社区、医院多方主体的协作管理。

ABSTRACT. The continuous evolution of digital information technology revolution accelerates the traditional medical and health industry into a new stage of digital health. In the new era, the "Internet + medical and health" mode promotes the optimization and upgrading of health management service quality, and comprehensively promotes the "active health management" mode, which focuses on improving comprehensive health management services, to take root. On the basis of clear definition of health management, through defining the connotation and characteristics of active health management service, this paper puts forward five ways to optimize the active health management service system, in order to point out a new development direction for the theoretical exploration, technology development and integrated demonstration of the construction of health China.

ABSTRACT. 在线医疗服务是缓解众多医疗问题的重要方式，研究旨在探讨用户在使用在线医疗APP的过程中会有哪些影响信任形成的因素，以及这些因素会对信任产生什么影响。研究采用各类在线医疗平台APP的用户评论文本，以信任源理论为基础，结合模糊数学理论，通过情感分析，从服务平台、服务主体、服务本身三方面出发，构建用户信任模型，挖掘文本中包含的信息。其结果能更好地服务于用户、医生、平台管理者。

ABSTRACT. Based on the Impellance-Instigation-Inhibition (I3) model, this paper identified the key factors affecting the behavior of the elderly in publishing knowledge-based short videos. Based on the analysis of 467 survey data, the findings of this paper are as follows: knowledge self-efficacy and perceived value as impellence, friend support as instigation significantly and positively affect the knowledge sharing intention of the elderly based on short video platforms, but the influence of family support is not significant. Through the mediation of knowledge sharing intention, knowledge self-efficacy, perceived value, and friend support indirectly promote knowledge-based short video publishing behavior of the elderly; while technology anxiety as inhibiton hinders the elderly's knowledge-sharing intention from transforming into knowledge-based short video publishing behavior.

ABSTRACT. 互联网医院井喷态势下，服务可及性是制约发展的关键因素，但已有文献对互联网医院服务可及性提高的实践路径缺乏充分的探讨。本文采用多案例研究法，通过对三种互联网医院发展模式的对比分析，总结出服务可及性的提高路径。研究发现，互联网医院可以从医疗服务的内容、质量和经济可及性三方面来提高整体服务可及性，因发展模式的不同其服务可及性提高的具体路径存在差异性。本文的研究结论为互联网医院的未来发展提供了启示和参考。

ABSTRACT. This paper adopts behavior experiment and structural equation model to reveal the influence of video features, namely scene cutting and presenter image, on users' knowledge learning effect and behavioral intention in health videos. Scene cutting and presenter image evoke the difference between argument quality and argument novelty. The main findings are: the impact of the argument valence and strength on intention to behave, the impact of argument novelty on learning effect, and the factors affecting intention to continuously use.

ABSTRACT. 保障“双碳”目标的顺利实现，应运数字化赋能低碳技术创新成为推进新发展理念的重要举措，更是实现文明跨越的创新抓手。基于“双碳”约束目标，构建以知识重构为中介变量、知识共享为调节变量的数字化转型驱动制造业低碳技术创新分析框架，实证探究数字化转型对低碳技术创新的影响。研究发现：制造业数字化转型对低碳技术创新产生显著正向影响，是推动低碳技术创新的重要动力来源；知识重构在数字化转型与低碳技术创新之间起到部分中介作用，发挥着关键桥梁作用；而知识共享对数字化转型与知识重构和低碳技术创新均起到正向调节作用。研究明晰了数字化赋能制造业低碳技术创新过程存在的差异性、适宜性，回答了数字经济时代如何有效利用知识管理实现数字化、帮助产业低碳创新升级问题，对引导产业绿色高质量发展具有重要参考意义。

ABSTRACT. 数字化时代，新型消费市场能在短短几年时间内从无到有，快速成长并形成市场影响力。新古典主义市场观难以解释当前这种新型消费产业的市场发育，而市场塑造的研究也大都聚焦核心企业的活动，忽视了小企业（市场主体）的作用。本研究综合了制度理论和社会涌现范式的理论视角，探究小企业如何通过合作、集体行动来塑造市场。本文选取山东曹县汉服产业市场发育案例作为样本，探析集体行动如何完成市场塑造的过程。结果表明：小企业集体合作塑造市场会经历四个阶段，“能人引领”阶段依靠个人特质推动互动交流；“促进联合”阶段涉及到有共同目标的主体，通过建立共同意愿，有选择地吸引同类市场参与者来解决共同问题；“建立合法性”阶段是集体活动通过谈判、制定规则和建立集体声誉等获得合法性，合作者网络也会随之发展；“形成新主流市场”阶段，正式制度和共同规范建立，集体可以利用市场影响力，对反对者和批评者采取相应市场手段，强化自我和侵蚀对手。数字技术是这一演化过程的重要催化因素。研究结论为小企业树立新型市场观，通过集体合作塑造新市场提供了理论和实践启示。

ABSTRACT. 数字经济时代，数字创新是助力组织获取竞争优势和提升组织绩效的战略举措。本文从数字创新四个方面出发，构建了对组织绩效影响的理论模型，在156份有效问卷的基础上，运用SPSS25.0和AMOS24.0对数据进行分析，试图揭示数字创新四个方面与组织绩效之间的路径关系。研究表明：数字组织创新、数字商业模式创新正向显著影响动态能力；动态能力在数字创新和组织绩效间发挥中介作用；在组织绩效提升中，数字创新和动态能力培育发挥着关键作用。

ABSTRACT. 数字支付在数字经济中扮演着重要角色。本文采用系统性综述法分析数字支付领域用户使用意愿对实际行为的解释和预测情况。通过在Web of Science数据库进行关键词搜索并根据纳入标准进行筛选，最终锚定了18个数字支付研究。本研究发现使用意愿与行为之间存在较大差异，这说明使用意愿对行为的预测与解释是不充分的。本研究进一步分析了这种差异产生的4种原因。揭示数字支付领域用户使用意愿对行为解释不足的现象对研究者具有一定启示性价值。

ABSTRACT. 在深化金融供给侧结构性改革，建设数字中国的背景下，金融科技赋能金融机构对释放实体经济发展活力,推动金融体系脱虚向实具有重要意义。本文采用2011-2020年31省市的面板数据，实证检验金融科技对金融支持实体经济及不同实体行业效率的影响，并基于中国银保监会官网金融许可证信息爬取的258772条金融机构数据，结合4998家上市公司的数据构建金融供求结构指标，从金融供求结构的视角考察金融科技脱虚向实的影响渠道。研究发现,金融科技推动了金融支持实体经济效率的提升，对金融支持第三产业效率的影响较为明显，各种稳健检验均支持这一结果。在金融供给端，金融科技从增强企业信贷可得性、加速金融机构本地化程度、促进金融机构多元化发展、提升金融可触及性和金融集聚效应五个维度有效弥补金融供给的不足；在金融需求端，金融科技通过深化实体行业与金融业关联程度，提高资源配置效率，从而改善金融供求结构的不平衡，提升金融支持实体经济的效率。最后，金融科技的脱虚向实作用在对外开放水平低、金融机构网点数量较少和创新水平较低的地区更为显著。本文的研究揭示，金融科技的应用为增强金融服务实体经济的能力，改善金融供给结构的不平衡提供了契机。

ABSTRACT. 本文采用模糊集定性比较分析（fsQCA）方法，以全国31省市1470家中小板和创业板企业为例，探索数字化转型和企业资本对中小企业绩效的联动效应。分析结果表明六个前因条件（数字化转型、物质资本、人力资本、创新资本、流程资本、关系资本）都不能构成中小企业绩效的必要条件；高水平中小企业绩效是多因素协同联动作用的结果，存在三条等效驱动路径；东中西部地区中小企业存在驱动高水平绩效的差异化路径。本文研究为中小企业实现高水平绩效提供一定的管理实践建议。

ABSTRACT. 为实现基于海量用户评论快速准确评价商家好坏。本文针对商家用户评论构建了外卖领域情感词典、主题词典以及商家情感指数模型，并以10个商家共计2000条数据进行实验。结果显示本文构建的词典和情感分析方法对情感极性识别准确度达97.5%，由此计算的情感指数值与商家综合评分整体趋势保持一致，表明在线评论情感指数可用于商家评分。本文提出的用文字代替打星的评分方法去除了空评和重复评论避免了控评行为，实现了对商家推荐服务的优化。

ABSTRACT. 本文基于不确定性降低理论，使用半结构式深度访谈法研究直播电商情境因素作用下在线评论对消费者退货意愿的影响，提出评论系统改进对策。研究表明：直播电商中消费者依然会阅读评论，但时间压力，群体压力和互动性分别通过降低信息处理能力，诱发从众效应和改变消费者的信息搜寻行为影响了在线评论降低产品不确定性方面的作用。直播电商平台可以通过弹幕展示可视化评论、增加评论搜索功能和定制化评论排序对评论系统进行改进。

ABSTRACT. Based on the perspective of explanation level theory and perceived value theory, this paper uses experimental research methods to explore the mechanism and boundary conditions of different forms of recommendation (text vs. short video) on consumers' purchase intention. The study found that compared with pictures and texts, short videos make consumers have higher purchase intention, and perceived value plays an intermediary role in the relationship between recommendation methods and purchase intention. Perceived psychological distance mediates the relationship between recommendation methods and perceived value. Bloggers' identity sources (plain people vs. celebrities) regulate the relationship between recommendation methods, perceived value and purchase intention. Experiential information disclosure regulates the relationship between recommendation methods and perceived psychological distance.

ABSTRACT. Based on the KMV model to assess credit risk , this paper explores the influence of senior executive changes on enterprise credit risk. On this basis, we can explore the adjustment effect of enterprise incentive mode and governance mode on this influence. The research results found that the credit risk of an enterprise increases significantly after the change of senior management, and the second position of senior management will strengthen the negative effect of senior executive change on the credit risk of enterprises, while the high concentration of equity of enterprises will weaken this negative effect. In addition, the research results found that both salary incentive and equity incentive can weaken the negative effect of senior executive changes on enterprise credit risk.

ABSTRACT. 本文用程序化扎根理论的方法，从半结构化的访谈中获得原始资料，根据各级编码、译码探究知识互动的关键影响因素，构建影响因素模型。用QCA的方法进行实证分析，以组态逻辑阐述影响因素的内在关联机制。研究发现5条能够产生较高行为意向的条件组合，将其归纳为互动平台主导型、平台和感知价值协同型、动机和障碍协同型3种构型。感知价值、互动动机、感知障碍和互动平台对产生混合教学模式下的知识互动行为意向有较强的推动作用。

ABSTRACT. In social Q&A community, users can exchange knowledge, which is also one of the ways for users to obtain knowledge. Based on social capital theory, this study explored some factors that affect users' continuous contribution in social Q&A community from the perspectives of structural capital, cognitive capital and relational capital. We collected data from 8,321 users on China's largest online community Q&A community platform and constructed a dynamic panel model. The results show that structural capital, cognitive capital and relational capital affect users' sustainable contribution behavior. In addition, we find that as an incentive, the badge plays a positive role in the process of structural capital's influence on users' sustainable use behavior.

ABSTRACT. 文章用内容分析法系统分析国内基于NVivo的质性分析研究成果以归纳基于NVivo的质性分析步骤及其应用模式，进而提升质性分析的研究及应用规范性、智能性及理论生成可追溯性。基于NVivo的质性分析步骤依次为研究者界定问题；按理论构建目的混合质性、量化方式收集资料并依托NVivo处理资料，依托NVivo进行资料内容的词频分析、编码分析、聚类分析并生成概念节点体系；依托NVivo按相关性进行资料词频、编码、聚类分析结果的关系分析、比较分析以建构理论或模型，并用编码信度、理论饱和度检验质性分析结果信效度。现有相关研究存在编码过程不够规范、信效度检验不够重视、理论贡献不够丰富等问题。应用模式由研究者根据问题、现象研究框架及内容清晰度，结合研究深度、广度选择运用描述类、探索类、整体类质性分析思路，以发现资料数据中变量关系模式、构建描述与诠释框架进而建构理论或模型。

ABSTRACT. 本文基于价值敏感性设计(VSD)，从结果悖论视角开展了一项深入的探索性研究以揭示用户使用在线医药O2O平台的重要价值，及其价值实现与否对用户体验的影响，并由此构建了在线医药O2O平台价值形成和影响的理论模型。首先，本文通过内容分析法发现用户使用平台主要关注人伦价值、技术伦理价值、社会伦理价值三个维度的十项价值，价值实现与否分别会让用户感受到四种不同形式的积极体验与消极体验；其次，在内容分析的基础上，本文进一步运用具有较强解释力的明确集质性比较分析法（csQCA）以探究平台价值与用户体验间的相关关系，研究发现单一技术伦理价值（可用性）、社会伦理价值（自主）实现或人伦价值（隐私或信任）与技术伦理价值（可用性）组合实现、技术伦理价值（可用性）与社会伦理价值（沟通、自主）组合实现会影响用户积极体验，而单一技术伦理价值（安全或透明）、人伦价值（隐私）、社会伦理价值（人类福利）未实现，人伦价值（希望）与社会伦理价值（公平）组合未实现、人伦价值（希望、信任）组合未实现会造成用户消极体验。

ABSTRACT. 在信息产品营销中，企业既可通过提供奖励，鼓励用户向潜在用户进行推荐，又可通过功 能研发，增加产品的内生网络外部性。本文基于一对一推广的博弈论模型，研究企业如何在产品端 的“病毒性设计”和市场端的“奖励推荐”策略之间进行平衡。结论表明，产品的病毒性设计与推 荐奖励策略之间存在复杂的联系:奖励分配方式由产品与市场特征决定;在奖励推荐人时，产品策 略和营销策略相对独立;而当奖励被推荐人时，二者存在替代关系。

ABSTRACT. 本文以启发式—系统式模型（HSM）为理论基础，基于某B2C平台客服聊天与交易的实际数据，会话分析视角下采用文本挖掘、情感分析等方法实现情感和交互变量的测量，通过logistic回归对消费者购买转化的影响因素进行分析。主要研究发现是：在启发式线索中，会话中的消息数量、表情符号对消费者购买转化存在正面影响，消息长度、客服响应时间对消费者购买转化存在负面影响；在系统式线索中，消极情感强度对消费者购买转化存在负面影响，积极情感强度对消费者购买转化没有显著影响。

ABSTRACT. 用户主要通过文本的方式为平台贡献创意，创意语言风格传达了重要信息，本研究关注文本语言风格对创意扩散度的影响。爬取典型开放式创新平台Thingiverse的数据，使用文本挖掘方法识别创意文本语言风格，探索创意文本语言风格对创意扩散度的影响，并使用机器学习算法对创意扩散度进行预测。结果表明，四种语言风格包括逻辑诉诸、人格诉诸、情感诉诸和信息熵均对创意扩散度产生正向影响，且创意认可度中介了语言风格对创意扩散度的影响。

ABSTRACT. 在线医生团队的多样性信息可能起到“双刃剑”作用，不同地影响患者。本文从多样性视角研究医生性别、科室及医院多样性对团队及其领衔专家绩效的影响。从领先健康平台上收集的2 588个团队相关数据，并进行实证建模分析。研究表明：性别和科室多样性分别正向和负向影响团队绩效，高级职称医生人数占比负向调节性别多样性的影响，这些多样性均正向影响领衔专家个人绩效。研究结果对于医生团队组建和在线服务绩效管理有理论和实践意义。

ABSTRACT. 互联网新闻已被很多研究用于物价指数的监测和预测。但在对于某一分类别物价的舆情指数构建过程中，经常会出现舆情观点新闻缺失的问题。在对没有舆情讨论的时点的舆情数据进行数据补齐的时候，不仅需要考虑该缺失数据时点前后的舆情值，同时也要考虑该时点其他相关主题的舆情值。为了更好地刻画不同相关主题下新闻舆情之间的复杂关联关系，本文提出了一种基于图神经网络自学习模型（AGNN）的舆情指数网络构建方法。这种方法将九个食品相关舆情指数作为网络节点，相应的量化后的日度舆情观点值作为时间序列，设计了一个图神经网络模型对存在的隐性图结构进行学习，构建成动态舆情指数网络，实现对于稀疏的细项新闻观点取值的补齐。为验证所提出的舆情指数网络模型的有效性，我们利用所提出的基于AGNN舆情指数网络的模型对各个食品分类相关的新闻观点数据缺失进行填补，并将其用于消费者价格指数食品项的预测。与其他数据补齐方式相比，我们经过我们提出的模型补齐的舆情数据对食品价格的预测精度提升最大。可见，本研究所提出的舆情指数网络模型能够更有效地刻画食品相关不同主题下，该方法对于其他类似的舆情指数集合建模问题也有一定的借鉴价值。

ABSTRACT. 本文的目的是在大数据框架下，讨论如何利用机器学习算法筛选公司财务舞弊与财务异常行为预警的特征指标的刻画及相关应用，即，我们首先筛选可描述公司财务舞弊与财务异常行为相关的关联“特征指标”，然后基于这些指标为基础，进行关联特征选取并针对其预警或解释能力进行甄别测试。具体来讲，通过刻画财务舞弊的特征指标和基于非结构化的公司治理结构的风险特征的提取，建立解读公司治理结构（特别是针对财务舞弊行为等方面）的风险预警体系需要的关联特征指标，并将这些指标应用于真实案例的测试分析。我们的案例分析表明，从财务到公司治理框架层面构建的针对财务舞弊行为和财务异常状态的特征指标刻画可以达到预测财务舞弊和预警的目的。本文的创新之处在于：基于公司舞弊三角原理，尝试采用大数据特征提取方法，能够构建刻画公司财务质量和公司治理结构的财务舞弊风险评估体系需要的关联特征因子指标； 同时我们也期待支持该体系的特征指标可以对公司可能要发生的财务舞弊进行及时的预警，从而促使行业健康发展并避免欺诈等不好行为带来的潜在损失.

ABSTRACT. Aiming at the heterogeneity of needs of various participants in the process of emergency rescue. In the process of multi cycle emergency material scheduling, the disaster satisfaction and government satisfaction are defined, and the multi cycle scheduling model of emergency materials aiming at the maximum comprehensive satisfaction is constructed. A random leapfrog algorithm based on reverse learning mechanism is designed to solve the model. The reverse learning mechanism and genetic crossover operator are embedded in the random leapfrog algorithm to improve the search ability of the algorithm. The superiority of the proposed algorithm is proved by four test functions, and the model is verified by an example. The example results show that the solution accuracy of the proposed algorithm is better than that of random leapfrog algorithm and genetic algorithm.

ABSTRACT. Information asymmetry in online consultation is widespread in online health communities (OHCs). As a key quality signal, it is very important to solve the problem of information asymmetry between doctors and patients. Basis on social capital theory, combining the relevant concepts in knowledge management and signal transmission theory to construct a research model, this study aims to explain the deep-seated mechanism of how doctors' popular science works affect patients' online consultation intention. By adopting machine learning and natural language processing technology, a new variable representing "common language" was extracted from the unstructured data of 5927 popular science articles and 43624 online consultation records, and its influence mechanism on online consultation intention was investigated. It is found that the common language of doctors' popular science works has a positive impact on patients' online consultation intention through the mediation of cognitive trust. Moreover, the professional qualifications of doctors can significantly moderate the impact of popular science article trust on patients' willingness to online consultation. The results of this study extend the theoretical connotation and interpretation scope of social capital theory, and provide important reference for doctors to provide popular science knowledge and efficient operation of online medical communities.

ABSTRACT. [Objective] To propose strategies to control the spread of rumors information and strengthen the spread of anti-rumor information by comparing the diffusion status of rumors and anti-rumor in social media. [Methods] Taking the incident of "Cao Moutao died of burns" as an example, we constructed Weibo users' retweeting relationship, calculate node influence with five network centrality indicators including degree center, and selected the SIR model to analyze the diffusion pattern of rumors and anti-rumor information. [Limitation] User infection rate and recovery rate should be adjusted according to individual situation. [Conclusion] Node influence is related to the closeness of this node and other nodes, identity attributes; Compared with anti-rumor information, rumor information has characteristics of explosiveness and single source.

ABSTRACT. Self-endorsement is the third way to increase credibility besides authority endorsement and identity endorsement, and it is widely used in the field of social media and short videos. The individual spends years on providing original, valuable, and quality content to gain audiences’ approval and support. This study creates a new model which is focused on testing the influence of self-endorsement-related factors, with the help of a questionnaire survey and data analysis. The results suggest that perceived interactivity, as the decisive factor, has a huge influence on WOM information credibility and information usefulness. Information usefulness is significant to the users’ intention to continue using the media platform or subscribe to the bloggers, whereas WOM information credibility is non-significant. However, compared to traditional media, social support is not of great concern to the audiences of short videos. This study creatively concludes the principles of self-endorsement, explaining its exertion and effect.

ABSTRACT. 本文根据注意力经济理论，采用回归分析方法，基于哔哩哔哩平台视频数据，分析了社交媒体平台上网络视频的信息娱乐化对注意力经济的广度、深度、参与度和效度中的不同维度的影响。实证研究结果表明，信息娱乐化特征中，正面和负面的极端情感对注意力经济具有显著影响，内容故事化具有显著的积极影响，明星人物、软新闻主题和“标题党”具有显著的消极影响。从在线视频和媒体平台的角度，时长碎片化对注意力经济具有显著影响，呈现方式多样化、标签数量、权威媒体具有显著的积极影响，话题数量具有显著的消极影响。

ABSTRACT. 以30个企业危机案例为样本，采用扎根理论研究方法，构建了企业危机在线传播中公众感知及行为反应形成机制的理论模型。研究发现：企业危机在线传播中，客观刺激、心理状态和行为反应是公众视角下的三大主要关系类别，公众信息行为遵循“刺激-心理-反应”的生成路径。具体地，危机属性、企业特征和企业响应作为客观刺激因素引发公众特定的心理状态，包括心理归因、期望和共情，而心理状态进一步影响公众的情绪反应和行为意愿。

ABSTRACT. 产品累积的在线评论内容是支撑消费者决策的重要信息来源，以往基于评论动机及评论有用性的产品在线评论生成机制研究忽视了消费者决策过程中不同阶段的信息价值。本文基于预测用户生成内容受欢迎程度的替代因素：内容新颖性，定义了评论内容新颖性；分析讨论消费者在决策过程中获取的产品公共信息、产品体验信息对评论内容新颖性的影响机制，明确了评论内容生成的信息来源。本文基于2007年至2012年消费者在京东商城发布的1,566,220条评论数据进行了实证分析，研究结果表明消费者获取的产品体验信息量正向影响消费者评论中不同于其已知公开信息的部分（即评论内容新颖性）。消费者对产品公共信息的了解程度负向调节产品体验信息量对评论内容新颖性的正向影响。且上述关系受到暴露新属性、暴露新观点的机制的交替影响。研究结果为商家和平台优化产品评论中产品信息的曝光程度提供重要管理启示。

ABSTRACT. The wide spread of misinformation on social media has caused huge threats to the physical and mental health of the public. It is of paramount importance to find effective intervention measures to reduce misinformation dissemination. We conducted a systematic literature review by analyzing 363 papers included in major academic databases. Our review is expected to (a) investigate the features of published research related to curbing the spread of misinformation on social media platforms, and (b) provide recommendations for future research on the topic.

ABSTRACT. 新兴技术为传统媒体的数智跃迁提供了强大支持，智能电视作为典型代表，是客厅经济的流量入口和强大载体。电视台等内容提供者一方面享受着客厅经济的红利，另一方面也面临着多重挑战，例如电视剧接档期间的观众流失。借助智能分析技术，本研究在智能电视中电视剧接档情境下，探究内容产品相似度与消费者留存之间的关系，并发现电视剧相似度与消费者留存存在倒U型关系，为客厅经济中电视台在内容产品的序列决策方面提供了借鉴。

ABSTRACT. 在线下影院遭受疫情冲击的当下，流媒体平台中的电影作品开始获得越来越多消费者的青睐，电影制作商、流媒体平台管理者也亟需理解消费者的电影偏好在线下、线上观看渠道中的差异。对此，本文以2011-2020年我国2122部院线电影为样本，分析流媒体平台相较于线下影院的消费者偏好变化，并探究两个渠道中电影份额分布的差异。研究结果表明：相对于线下影院，在流媒体平台上风格轻快的电影会获得更多观看，而演员和导演号召力的作用下降；此外，流媒体平台上的电影份额分布呈现出更明显的长尾现象，这一结果证实了长尾理论在数字内容产品中的适用性。