近年来,城市化进程的不断加剧以及机动化趋势的日益显著,推动了城市人口的不断聚集以及城市范围的快速扩张。然而,城市无序扩张过程也带来了一系列问题,如郊区产业化滞后、用地功能单一以及公共交通设施滞后等,不仅造成了城市周边居民的通勤距离和通勤成本的增长,也加重了现有交通设施的负担,加剧了城市中心区域的拥堵。城市的通勤拥堵严重影响城市的社会经济发展,并逐渐成为世界各国所共同面临的社会问题。美国Inrex公司指出,2019年世界最为拥堵的十大城市集中位于拉丁美洲和欧洲,其中哥伦比亚首都波哥大的平均通勤时间达到191小时,成为了全球最为拥堵的城市。根据普查报告,2019年美国由于交通拥堵而导致的人均额外通勤时间达到99小时,所造成的直接和间接经济损失高达880亿美金。我国面临的城市拥堵问题也尤为严重,2014年中国北京、上海、广州、深圳居民的平均单程通勤时间为40分钟,超过世界平均水平（32分钟）。而根据IBM对全球各国主要城市的调查,北京和深圳的通勤痛苦指数分别位于全球第二和第三位。因此,如何更好的理解拥堵发生的机制并控制通勤时间的增长,成为了经济学和交通学研究人员所面临的挑战。当前的研究集中于对短期社会经济因素的探讨,却忽视了影响城市长期发展的基本地理结构,因而缺乏一致性的结论,对城市交通政策的参考价值有限。根据美国人口普查数据,平均通勤时间最长的前10个城市有8个都位于东西海岸（大型地理障碍）附近,而这些城市的地理结构由于地理障碍的存在也普遍较为复杂。然而,城市地理结构与通勤之间的联系并未被以往的研究讨论,地理障碍对城市地理结构的影响也没有被很好的刻画。通勤研究急需从地理学的角度出发,探索城市基本地理要素,尤其是地理障碍,对于个体交通行为以及宏观交通拥堵的影响。因此,为了填补城市通勤交通研究中的这一空白,本文利用美国的交通数据,从微观、中观、宏观三个层面验证了地理障碍对个人通勤行为、局部拥堵以及城市整体通勤效率的影响,并深入地发掘了地理障碍对通勤的作用机制。本文的主要研究工作如下:（1）研究了地理障碍对个人通勤活动的影响,从微观上发现了地理障碍对居民个人通勤行为的作用机制。通过实施空间划分和通勤点选取策略,建立地理障碍率刻画指标以及通勤轨迹与地理障碍的关联模型,验证了地理障碍的存在会对早高峰居民的基本通勤指标,通勤距离、通勤速度以及通勤时间产生影响。初步地从个体层面发现了地理障碍对个人交通行为的作用机制,为建立城市地理结构与城市拥堵之间的关系奠定了基础。（2）研究了地理障碍对城市局部拥堵以及通勤流的影响,发现地理障碍会增加城市中心区域的通行轨迹,降低地理障碍附近区域的拥堵疏散能力,以及减少通行方向的通勤流。本文通过构建市中心通行率指标、最短路径靠近海岸线比例,并基于波士顿居民职住分布数据,构建了有效通勤者与有效通勤流计算模型,发现地理障碍对个人通勤轨迹的影响,会使得市中心5km范围内的总体通勤轨迹数增加13.87%。通过引入物理学热传导方程,发现地理障碍的存在会迫使沿海区域的通勤者可选择的通行路径减少,从而降低了沿海区域的拥堵疏散能力,造成其长期的交通阻塞。发现城市地理结构会对局部通勤流产生影响,通行方向上每1%的地理障碍存在,会造成0.21%的通勤流量减少。上述结论进一步地发现了地理障碍对城市局部拥堵的作用机制,从中观尺度建立了个人通勤行为与城市局部拥堵之间的联系。（3）研究了地理障碍对城市总体通勤效率的影响。基于美国868个核心统计区（CBSA）的地理与社会经济数据,分别从地理学与城市经济学的角度出发,设计了地理指标的双重形态刻画指标,并建立了地理障碍形态指标与城市通勤效率之间的关联,发现高比例的地理障碍会增加城市居民平均通勤时间,使通勤者的出发时间提前,而地理障碍的复杂形态会加剧这种影响。上述结论最终验证了地理障碍的存在会对城市的总体通勤效率造成负面影响,对于城市交通布局优化方案的制定具有普适性强的参考作用。

随着社会化购物行业的高速发展，直播购物俨然成为企业销售新的增长点，企业如何选择直播购物主播类型成为企业关注的重点。由于直播购物具有社交性、即时性和虚拟性，消费者在直播购物中会更依赖于直播主播这一外部线索去评估产品，制定购物决策。且消费者在进行不同类型产品的购买决策时的价值驱动也有所不同。因此，本研究基于社会影响理论，探讨了直播购物主播类型（企业主播与名人主播）与产品类型（实用品与享乐品）的交互作用对消费者购买意愿和行为的影响。通过情景实验和二手数据发现，名人主播推荐介绍享乐品，促进触发消费者的认同机制，并提高其购买意愿和产品销量。企业主播推荐介绍实用品，促进触发消费者的内化机制，并提高其购买意愿和产品销量。本文拓展了社会化购物中对于直播主播类型的研究范畴，深化了对社会影响理论动力驱动的理解，同时对企业的直播主播选择策略具有一定的指导意义。

We propose a conceptual framework that includes the antecedents and consequences of firms' adopting and integrating robotics into their customer service operations. Drawing insights from literature on customer service, technology marketing, and computer science, our proposed framework elaborates on the concept of the degree of robotics adoption (DRA) as well as the antecedents (employee acceptance of robots and customer acceptance of robots) and multiple sequential consequences (service quality, customer long-term performance, and customer engagement) of DRA. We also discuss how the nature of the firm (Business to Consumer versus Business-to-Business, i.e. B2C vs. B2B), service characteristics (utilitarian vs. hedonic), and brand positioning (low equity vs. high equity) might moderate the relationship between DRA and service quality. Further, we provide actionable guidance for managers to adopt and integrate robotics into their customer service operations.

PurposePrevious studies on social influence and virtual product adoption have mainly taken users' purchase behavior as a dichotomous variable (i.e. purchasing or not). Given the prevalence of competing versions (basic vs upgraded) of a virtual product in online communities, this paper investigated the differences in the effect of social influence on users' adoption of basic and upgraded choices of a virtual product. It also examined how the effect varies with users' social status and user-level network density.Design/methodology/approachA natural experiment was conducted in an online game community. Two competing versions (basic vs upgraded) of a virtual product were provided for in-game purchase while a random set of users selected from 897,765 players received the notification of their friends' adoption information. A competing-risk model was used to test the hypotheses.FindingsSocial influence exerts a stronger positive effect on users' adoption of the upgraded virtual product than of the basic virtual product. Middle-status users have the greatest (least) susceptibility to social influence in adopting the upgraded (basic) virtual product than low- and high-status users. User's network density enhances the effect of social influence on adoption of both virtual products, even more for the upgraded one.Originality/valueThis research contributes to the social influence and product adoption literature by disentangling the different effects of social influence on basic and upgraded versions of a virtual product. It also identifies the boundary conditions that social influence works for each version of the virtual product.

Traditional congestion control protocols may fail to achieve consistently-high performance over a wide range of networking environments as their hardwired policies are optimized over specific network conditions. In this paper, we depart from conventional wisdom and propose GLIDER, a new congestion control protocol that uses deep reinforcement learning to be more versatile and adaptive to dynamic environments. In particular, GLIDER uses a framework based on Deep Q-Network, that a sender keeps adapting its congestion control strategies by continuously interacting with the network environment. In addition, the sender constantly sends data, making it challenging to apply reinforcement learning algorithms that require step-by-step state computation to congestion control. Therefore, we design a Dynamic Bisection Division Algorithm (DBDA) to discretize the packet transmission process into steps to ensure GLIDER'S feasibility on congestion control. We have used an extensive array of experiments on Pantheon to show that GLIDER can adapt well to varying buffer sizes and is resilient to random loss. Moreover, on wide-area inter-data center links, it can achieve 6.4x and 1.4x higher throughput than TCP CUBIC and BBR, respectively, and comparable performance as other learning-based congestion control protocols in the literature.

The product recall is becoming a challenging issue in supply chain management, and both suppliers and manufacturers are motivated to exert recall efforts to improve supply chain performance. Considering a supply chain composed of one supplier and one manufacturer with product recalls and demand uncertainty, we investigate the contract design by integrating the revenue sharing contract with three cost sharing policies, namely, fixed-rate cost sharing (Contract F), linear cost sharing (Contract L), and threshold-based cost sharing (Contract T), to coordinate the production quantity and recall efforts. Results show that cost sharing could improve the product recall probability and supply chain profits, but only Contract T can coordinate the supply chain in both production quantity and recall efforts. Moreover, the cost sharing rate in Contract T first decreases and then increases as the recall cost increases. When the recall cost is small (large), Contract T (Contract F) is optimal for the manufacturer. In contrast to our intuition, we find Contract L could overmotivate supply chain members to exert a recall effort that is greater than the first-best when the cost coefficient of recall efforts is large enough. However, demand uncertainty will mitigate the impact of product recalls on motivating recall efforts.

Mobile edge computing(MEC)is a promising technology for the Internet of Vehicles,especially in terms of application offloading and resource allocation.Most existing offloading schemes are sub-optimal,since these offlo

When powered by artificial intelligence (AI), the effectiveness of advertising generally improves. However, evidence shows that personalized recommendations in charity advertising may have a dark side. The existing literature about the effects of personalized recommendations in advertising is rooted primarily in outcome utility, including outcome benefits and costs. Nevertheless, consumers tend to sacrifice their own interests without expecting anything in return given that they cannot directly monitor and measure the behavior outcome in charitable consumption; this elicits in them a focus on their autonomy and signal utility in responding to charity advertising. Thus, in our article, we focus on the reasons that personalized recommendations have negative effects in charity advertising based on self-determination theory. Through five studies, the results reveal that consumers display lower donation intentions when they receive charity advertising with (versus without) personalized recommendations due to a decrease in perceived autonomy. In addition, this negative effect can be mitigated by servant communication styles and providing consumers with free choices. These conclusions not only enrich the literature on personalized recommendations in advertising, charity advertising, and AI marketing but also provide some guidance for advertisers to enhance the performance of personalized recommendations in charity advertising.

The commuting cost has increased inevitably under the background of industrialization and urbanization, which has vital impacts on daily life. In this paper, we implement an empirical study to explore the impact of commuting time on hedonic consumption and the underline mechanism of perceived stress based on the data of CFPS 2014 (a national social tracking survey project). The Tobit regression combined with the Heckman two-stage model is applied to correct the sample selection bias. The results show that the increase in commuting time will aggrandize residents' tendency and expenditure of hedonic consumption. Moreover, we conduct situational experiments as a supplement to verify the main effect and mediation mechanism to exclude alternative explanations. The mechanism analyses prove that the perceived stress caused by long commuting time is a possible mediator for hedonic consumption. This study enriches the relative studies on social influences of commute, especially proving that this social issue may provide some benefits for business practice.