Electronic Commerce Research and Applications最新文献

{"title":"Modelling greenhouse gas emissions of Q-commerce vs E-commerce logistics operations: A simulation and conceptual framework","authors":"Jay Daniel,&nbsp;Jehan Zaib,&nbsp;Harshal Pandurang Gund","doi":"10.1016/j.elerap.2026.101640","DOIUrl":"10.1016/j.elerap.2026.101640","url":null,"abstract":"<div><div>The rapid expansion of E-commerce and ultra-fast Q-commerce grocery services has heightened concerns about the environmental sustainability of last-mile logistics. Although previous research identifies last-mile delivery as a significant source of greenhouse gas (GHG) emissions in online grocery fulfilment, there is limited empirical analysis systematically comparing the emission profiles of E-commerce and Q-commerce models. This study addresses this gap by quantifying and comparing GHG emissions for both fulfilment systems through an integrated analytical and simulation-based methodology. Deterministic mathematical models estimate emissions across upstream and downstream logistics stages, while Monte Carlo simulations capture variability in delivery distances, vehicle utilisation, and order characteristics. Q-commerce emissions are benchmarked against established E-commerce emission values from secondary data to facilitate comparative assessment.</div><div>The results indicate that GHG emissions per item are substantially higher in Q-commerce (0.626 kg CO₂ eq) compared to E-commerce (0.252 kg CO₂ eq). Last-mile logistics account for approximately 99.46 per cent of total emissions in Q-commerce and 80.16 per cent in E-commerce, confirming downstream delivery as the primary emission hotspot. These findings demonstrate that ultra-fast fulfilment models impose significantly greater environmental burdens than conventional E-commerce and underscore the necessity for structured organisational strategies to mitigate emissions associated with rapid delivery operations.</div><div>In response, this study introduces the GHG Emission Strategy (GES) framework, grounded in the Input-Process-Output model, to link emission drivers, strategic levers and sustainability outcomes. The framework guides retailers in reducing Q-commerce emissions while extending the theoretical understanding of systematic carbon management and sustainable last-mile logistics across retail fulfilment systems.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101640"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149140657","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Personalized travel package recommendation algorithm based on improved variable neighborhood search","authors":"Chao Huang,&nbsp;Xian Li,&nbsp;Yifan Zhang,&nbsp;Yuzi Zhou","doi":"10.1016/j.elerap.2026.101647","DOIUrl":"10.1016/j.elerap.2026.101647","url":null,"abstract":"<div><div>As the tourism market expands and tourist demands diversify, personalized tourism recommendation systems are essential for enhancing tourist experience and satisfaction. Traditional travel package recommendation methods often focus solely on recommending the most preferred options to tourists, thereby prioritizing recommendation precision over diversity. They also often neglect factors such as the stochastic nature of travel time caused by varying traffic conditions. This paper addresses these issues by proposing a two-stage personalized travel package recommendation framework, named TPRIVNS. In the first stage, we propose an improved variable neighborhood search algorithm (QT_VNS), based on quality score and topic diversity, to construct a candidate pool of travel packages. The algorithm introduces two penalty factors (one for attraction quality score and one for topic similarity), and defines five neighborhood operators for package adjustment and optimization, with the objective of balancing average attraction quality and intra-package diversity. In the second stage, we propose a stochastic optimization model considering inter-package diversity for travel package recommendation, which incorporates chance constraints to handle stochastic travel time and is solved using the Sample Average Approximation (SAA) method. Finally, we evaluate the proposed TPRIVNS framework using real tourism data, assessing the method’s precision, diversity, and time reliability.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101647"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149199621","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI-generated summaries in E-commerce: a dual-process perspective on AIGS type and anthropomorphism as a moderator","authors":"Yanxi Tao,&nbsp;Chundong Zheng,&nbsp;Jiehang Song","doi":"10.1016/j.elerap.2026.101644","DOIUrl":"10.1016/j.elerap.2026.101644","url":null,"abstract":"<div><div>E-commerce platforms have increasingly adopted AI-generated summaries (AIGS) to facilitate consumers’ decision-making. Unlike traditional tools, AIGS leverage AI’s capability to synthesize extensive user-generated review data. However, how the presence and type of AIGS impact consumers’ decision-making in e-commerce remains underexplored. We classified AIGS into attribute-based AIGS and experience-based AIGS. Grounded in Dual-Process Theory, we conducted four randomized experiments (total <em>N</em> = 813) across varied consumer decision contexts. The results indicate that AIGS do not uniformly enhance consumer responses; rather, their effect varies by type: while attribute-based AIGS reliably improve responses through lower perceived cognitive effort, experience-based AIGS improve responses through perceived emotional value only when accompanied by high anthropomorphic cues. Theoretically, this study advances AIGS research by examining the effectiveness of AIGS based on their content-level heterogeneity, extending Dual-Process Theory to AIGS, and identifying anthropomorphism as a boundary condition. Practically, the findings offer actionable insights for platforms to tailor AIGS types and anthropomorphic cues to decision contexts.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101644"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149140659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Commercial poisoning in AI search: Economic incentives and vulnerability persistence in retrieval-augmented generation","authors":"Yuxuan Zhu,&nbsp;Wan Mohd Hirwani Wan Hussain","doi":"10.1016/j.elerap.2026.101629","DOIUrl":"10.1016/j.elerap.2026.101629","url":null,"abstract":"<div><div>AI search is increasingly becoming an interface through which consumers discover and compare products, and the retrieval layer supplying these systems is open to paid manipulation that carries no disclosure. This study terms such manipulation <em>commercial poisoning</em> and assesses it across three major providers (Google, OpenAI, Anthropic) in controlled experiments comprising 1800 observations. Six economy-tier models were evaluated over two cohorts using a synthetic corpus of consumer product content with contamination varied from 0 to 5%. Economy-tier models prove highly susceptible: 1% contamination produces bias rates of 50%–100%, and susceptibility rises sharply with query specificity, peaking on the product comparisons consumers rely on most. Contamination effects are governed by a threshold rather than a dose, so minimal investment secures near-maximal influence. Contextual poisoning circumvents explicit neutrality instructions in RLHF models at rates of 47.50% to 66.38%, against 4.17% under Constitutional AI (Cohen’s <span><math><mrow><mi>h</mi><mo>=</mo><mn>1.11</mn></mrow></math></span> to 1.49). Testimonial content consistently outperforms fabricated statistics in this retrieval setting, supporting the narrative-transportation account under AI-mediated product discovery. The proposed Commercial Poisoning Framework attributes these vulnerabilities to attacker economics that make manipulation self-financing, provider underinvestment arising from externalized security benefits, and the asymmetry between per-campaign attack and per-query defence. The findings identify commercial poisoning as undisclosed advertising that circumvents the separation between paid placement and organic content on which consumer discounting depends, distorts competition in favour of firms investing in retrieval optimization irrespective of product quality, and concentrates exposure among price-sensitive consumers and merchants.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101629"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148936504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DTCAR: Dual-view temporal contrastive alignment for recommendation","authors":"Zhiqiang Zhang,&nbsp;Tingting Feng,&nbsp;Jiangzhou Deng,&nbsp;Jianmei Ye,&nbsp;Huanyu Chen,&nbsp;Yong Wang","doi":"10.1016/j.elerap.2026.101634","DOIUrl":"10.1016/j.elerap.2026.101634","url":null,"abstract":"<div><div>To address the limitations of existing recommender systems in semantic expressiveness and dynamic interest modeling, we propose a dual-view temporal contrastive alignment recommendation model called DTCAR. DTCAR is designed to effectively integrate global collaborative signals and fine-grained semantic intents within a unified temporal framework. Specifically, users’ historical interactions are decoupled into a collaborative view and a semantic view. The collaborative view leverages a light graph convolution network to model high-order user–item relationships, while the semantic view employs a dedicated text encoder, LLM2Vec, to extract explicit user intents from reviews. For each view, we adopt parameter-independent yet structurally consistent time-aware self-attention encoders with exponentially-spaced time binning to model dynamic evolution along a shared timeline. To further align heterogeneous views, we introduce a bidirectional cross-level contrastive learning mechanism with symmetric NT-Xent loss, combined with adaptive gated fusion and residual connection to obtain unified user representations. Experiments on four public benchmark datasets demonstrate that our model outperforms state-of-the-art baselines. Compared with the second-best performing baseline, DTCAR achieves relative improvements of 5.98%–12.29% in NDCG@3 and 6.34%–11.81% in HR@3 across datasets. Moreover, a case study further highlights its interpretability in capturing users’ dynamic interest evolution.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101634"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148936501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Diversified service recommendation method based on cross-view contrastive learning","authors":"Jiawei Lu,&nbsp;Wanchuang Cai,&nbsp;Duanni Li,&nbsp;Linrong Jiang,&nbsp;Gang Xiao","doi":"10.1016/j.elerap.2026.101632","DOIUrl":"10.1016/j.elerap.2026.101632","url":null,"abstract":"<div><div>With the rapid development of service-oriented computing and cloud computing, the increasing number of Web application programming interfaces in various service sharing communities enables software developers to efficiently create composite functions of their interest. However, the abundance of functionally similar APIs makes it difficult to capture subtle functional discrepancies, thereby reducing composite service development efficiency and diminishing developers’ experience. Moreover, most existing service recommendation methods tend to focus more on recommendation accuracy while ignoring diversity, thus leading to monotonous recommendation results. To address these challenges, we propose a diversified service recommendation based on cross-view contrastive learning (DSR-CCL). Specifically, we employ a bundle recommendation strategy to model service scenarios and construct both single-service and bundle-service views. Graph neural networks are then used to learn representations across views, with customized propagation and aggregation mechanisms designed for heterogeneous node types. Furthermore, we propose data augmentation methods of edge dropping, node dropping, and feature shuffling, along with corresponding positive-negative sampling strategies, and adopt the InfoNCE loss to realize cross-view contrastive learning. Experimental results on real-world service datasets demonstrate that DSR-CCL achieves competitive recommendation accuracy while significantly enhancing result diversity. The source code and datasets are publicly available at <span><span>https://github.com/viivan/DSR-CCL</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101632"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004578","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"When to share content to rival platforms?","authors":"Hongyan Xu,&nbsp;Nengyue Zhu,&nbsp;He Huang,&nbsp;Yifan Dou","doi":"10.1016/j.elerap.2026.101638","DOIUrl":"10.1016/j.elerap.2026.101638","url":null,"abstract":"<div><div>This research explores the content sharing strategy of competing music platforms offering online music services. We address two issues: the distribution of exclusive copyrights and rival procurement of licensed copyrights. The distribution generates the effects of <em>copyright sharing</em> and <em>copyright acquiring</em>. We find that under the scenario where only one platform can distribute exclusive copyrights, a unilateral distribution equilibrium, in which one platform distributes copyrights and the rival procures licensed copyrights, arises when the distribution share is low and the base payout rate is moderate, or when the distribution share is high and the base payout rate is not high. By contrast, under the scenario where both platforms can distribute exclusive copyrights, a bilateral distribution equilibrium, in which both platforms distribute copyrights and procure licensed copyrights from each other, arises when the copyright value is low and the base payout rate is high, or when the copyright value is not high and the base payout rate is low. Moreover, a higher distribution share of exclusive copyrights or stronger network effects may hinder the unilateral (bilateral) distribution equilibrium under the scenario where only one platform (both platforms) can distribute copyrights. Additionally, unilateral or bilateral distribution does not necessarily bring higher social welfare. For consumers, bilateral distribution is more favorable than unilateral distribution.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101638"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149043667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on green supply chain decision-making under the collaboration of blockchain technology and digital transformation","authors":"Guibao Liang,&nbsp;Yulong Dai,&nbsp;Qiao Xia","doi":"10.1016/j.elerap.2026.101635","DOIUrl":"10.1016/j.elerap.2026.101635","url":null,"abstract":"<div><div>Under the background of the global “dual-carbon” target, the green transformation of supply chains faces two bottlenecks, namely the high cost of green research and development, and the bullwhip effect caused by supply chain information asymmetry. Therefore, this study considers the collaborative impact of blockchain technology and manufacturers’ digital transformation on the green supply chain, constructs a supply chain model consisting of manufacturers and retailers, and sets up four decision-making scenarios through the Stackelberg game, and finds that: (1) The digital transformation can enhance the greenness of products, and blockchain can reduce greenness while mitigating the bullwhip effect; (2) Government subsidies can positively regulate greenness, especially hedging against the inhibiting effect of blockchain technology on greenness; (3) Dual technology applications can significantly optimize supply chain costs, but have lower greenness than single digital transformation scenarios due to the negative impact of blockchain; (4) Consumer greenness preference determines the benefits of the technology, with demand driven by greenness when preference is high, and price playing a dominant role in the case of low preference.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101635"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148936505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Metaverse for e-commerce excellence: Role of digital capabilities and organizational culture in driving marketing performance","authors":"Shaofeng Wang,&nbsp;Hao Zhang","doi":"10.1016/j.elerap.2026.101637","DOIUrl":"10.1016/j.elerap.2026.101637","url":null,"abstract":"<div><div>The Metaverse is rapidly reshaping the landscape of electronic commerce, offering novel avenues for firms to engage customers and enhance marketing strategies. However, understanding how to effectively leverage this emerging digital environment remains a critical challenge. This study empirically investigates the impact of firm-level Metaverse competencies—comprising infrastructure, business spanning, and proactive stance—on digital marketing capabilities (information management, planning, and implementation) and subsequent marketing performance. Furthermore, we examine the crucial moderating role of digital culture, which we define in five facets and delineate both conceptually and empirically from data-driven culture, in amplifying these relationships. Employing a mixed-methods approach that combines a temporally separated survey of 188 firms analyzed with PLS-SEM and fsQCA, a nine-month follow-up wave supporting an autoregressive cross-lagged test on 112 of those firms, and 18 systematically coded semi-structured interviews with executives actively engaged in Metaverse marketing, our findings reveal that Metaverse competencies significantly bolster digital marketing capabilities, which in turn drive superior marketing performance. Notably, a robust digital culture strengthens the positive effects of Metaverse competencies on digital marketing capabilities, whereas a data-driven culture does not display an equivalent amplifying effect in the follow-up subsample. This research contributes to the e-commerce literature by providing a nuanced framework for firms to develop and deploy Metaverse strategies effectively, offering actionable insights for managers to achieve competitive advantage and optimize marketing outcomes in this transformative digital frontier.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101637"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Is high-arousal language always persuasive? The role of congruence between verbal and nonverbal cues in virtual live streaming","authors":"Xiaoli Lu,&nbsp;Linlin Xu","doi":"10.1016/j.elerap.2026.101639","DOIUrl":"10.1016/j.elerap.2026.101639","url":null,"abstract":"<div><div>Live-streaming e-commerce increasingly relies on virtual hosts to deliver emotionally resonant product recommendations. Previous research indicates that high-arousal language enhances persuasiveness, yet in practice, it does not always translate into consumer purchase intention. Based on congruence theory and the persuasion knowledge model (PKM), this study examines how congruence between virtual streamers’ verbal and nonverbal cues at the level of emotional arousal influences consumers’ purchase intention, as well as the underlying psychological mechanism and boundary condition. Across four experiments, the findings show that high-arousal language effectively enhances purchase intention only when it is supported by congruent facial and bodily expressions. In contrast, when the two types of cues are incongruent, consumers experience stronger cognitive conflict, which increases perceived manipulative intent and thereby weakens the persuasive effect. The core findings are replicated across different product categories and virtual streamer styles. In addition, this effect is more pronounced among consumers with lower imagery difficulty. This article demonstrates that the persuasive efficacy of high-arousal language critically depends on its coordination with nonverbal cues. It advances understanding of multimodal persuasion in virtual live-streaming environments and highlights the importance of maintaining dynamic emotional consistency in virtual host design.</div></div>","PeriodicalId":50541,"journal":{"name":"Electronic Commerce Research and Applications","volume":"79 ","pages":"Article 101639"},"PeriodicalIF":6.8,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149043666","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}