Search Costs Research Articles

Contrastive Learning for Enhanced Retrieval on Online Medical Platforms

Online medical platforms have rapidly developed in recent years. Due to search inaccuracies and incomplete information provided by doctors, some suitable doctors are excluded during the first-stage retrieval. To enhance doctor retrieval and reduce user search costs, obtaining disease word embeddings is essential. Existing methods primarily rely on external corpora and sentence context, which may not align with downstream tasks, and many languages lack standardized medical corpora. Therefore, we integrate contrastive learning with medical knowledge to generate and augment data using simulations and propose a simple network structure for training on the constructed samples. We created two task datasets based on the platform’s data. Experimental results demonstrate that our framework achieves superior outcomes with lower embedding dimensions. In the similarity task, our framework attains an accuracy of 0.854, and in the retrieval task, it achieves an F1 score of 0.853, surpassing the current best results. Our framework has been successfully implemented on Baidu Health, one of the China’s largest online medical platforms, serving over 10 million users. This framework effectively simulates a doctor retrieval system, optimizing the process to ensure more accurate and comprehensive retrieval of suitable doctors.

Hessian-based mixed-precision quantization with transition aware training for neural networks.

Model quantization is widely used to realize the promise of ubiquitous embedded deep network inference. While mixed-precision quantization has shown promising performance, existing approaches often rely on time-consuming search process to determine the optimal bit configuration. To address this, we introduce Hessian-based Mixed-Precision Quantization Aware Training(HMQAT) to decrease the search overhead of bit configuration. By using the sensitivity metric of joint average Hessian trace and parameter size, HMQAT effectively guides the search process. We solve the optimization problem about bit configuration automatically using a Pareto frontier method. The lowest search overhead can be achieved through our scheme. Additionally, our approach incorporates quantization transition aware fine-tuning of scale factor. This strategy consistently ensures optimal inference performance along the accuracy-size Pareto frontier across multiple models. We extensively evaluated our method on ImageNet and CIFAR10. In particular, we show that compared to the baseline, HMQAT achieves a 10.34× reduction in model size while retaining 99.81% of the Top-1 accuracy on ResNet18 for ImageNet. Moreover, HMQAT surpassed the state-of-the-art mixed-precision quantization methods in compressing neural networks with reduced search cost and achieving a satisfying trade-off between size and accuracy. This study paves the way of deploying neural networks on lightweight devices.

Choice Overload and the Long Tail: Consideration Sets and Purchases in Online Platforms

Problem definition: This paper examines frictions in the shopping funnel using empirical clickstream data from an online travel platform. We analyze (a) customers’ heterogeneous search and purchase behaviors and (b) their reactions to changes in assortment size. We then develop a consider-then-choose model to generalize our findings. Methodology/results: We characterize the online customer journey as a two-stage consider-then-choose framework. In the consider stage, we analyze the consideration set formation and show that heterogeneity—familiarity with the assortment—amplifies the number of options; in the purchase stage, it drives preferences for niche versus popular choices. A real-world high-stakes field experiment reveals that shrinking the menu produces mixed results: highlighting the market for the long-tail for some customers and reflecting choice overload for others. Finally, we build a psychologically rich consider-then-choose model with (a) heterogeneous preferences for product features and (b) heterogeneous search costs moderated by search fatigue, theoretically characterizing the impact on consideration sets and conversion rates. Managerial implications: Identifying frictions in the shopping funnel is critical for online platforms, especially when pain points hurt click-through or conversion rates. Which options matter to which users? What is the right assortment size? Although online platforms can offer virtually unlimited assortments, managers may assume frictionless environments—which is not always the case. Our findings offer insights into improving the customer journey by considering heterogeneous preferences and boundedly rational heuristics. Supplemental Material: The online appendices are available at https://doi.org/10.1287/msom.2021.0318 .

Fashioning Authenticity: How Digital Product Passports Could Shape the Future of Trademark Law

Abstract In the evolving landscape of the fashion industry, the interplay between emerging regulatory frameworks, such as the Digital Product Passport (DPP) Regulation, trademarks and narratives about authenticity, is critical in shaping consumer purchasing decisions. This article investigates how marketing, legal and technology might shape consumer behaviour in the future and highlights what legal issues might arise. Through a literature review, an analysis of the current DPP framework, the examination of recent market initiatives, and a case study, this article suggests that DPPs could reduce the importance of trademarks in the marketplace. Indeed, by providing consumers with verifiable digital reference points tied to authenticity narratives, DPPs would lessen the importance of trademarks for reducing consumers’ search costs, thus challenging its economic foundations. Findings underscore the growing influence of technology on consumer choices, the need for trademark experts to reassess their roles in crafting brand narratives, and the growing need to redefine the foundations of trademark law. Ultimately, this article aims to provide actionable insights for recalibrating trademark protections to foster a more innovative fashion industry.

A Revealed Preference Approach to Identification and Inference in Producer-Consumer Models

This article provides a new identification result for a large class of models in which consumers participate in production. I show that consumer preferences are necessary and sufficient to identify production functions through cross-equation restrictions implied by first-order conditions. In addition, I derive a nonparametric revealed preference characterization of the class of models that exhausts its empirical implications. Finally, I use a novel and easy-to-apply inference method that is valid under partial identification. This method can be used to statistically test the model, can deal with any type of latent variables (e.g., measurement error), and can be combined with standard exclusion restrictions. Using data on shopping expenditures and shopping intensity from the NielsenIQ Homescan Dataset, I show that a doubling of shopping intensity decreases prices paid by about 15 % . At the same time, I find that search costs are significant, hence, largely diminishing benefits of price search.

The Power of Brand Signature: Unpacking its Impact on Customer Loyalty

In the contemporary era, consumer preferences are rapidly evolving towards branded products, leading to a stronger connection between consumers and markets. Marketers are using social media more and more to promote their brand’s signature. This paper delves into the concept of brand signature and its impact on customer loyalty. Brand signature is key to brand identification as it represents a brand uniquely through elements, like a logo, design, or tagline, that uniquely identifies and represents a brand. These components work together to create a strong and lasting brand identity, enabling easy recognition by consumers. A well-defined brand signature leads to stronger customer relationships and loyalty. A strong brand signature minimizes customer search costs, enhances brand recognition, and foster customer loyalty by reinforcing the brand’s identity. As a result, brands prioritize their brand signature to enhance customer loyalty, achieve better market differentiation, and enhance brand value.

Pandora’s Problem with Combinatorial Cost

Pandora’s problem is a fundamental model in economics that studies optimal search strategies under costly inspection. In this paper, we initiate the study of Pandora’s problem with combinatorial costs, capturing applications where search cost is nonadditive. Weitzman’s celebrated algorithm (1979) demonstrates that for additive costs, the optimal search strategy is nonadaptive and computationally feasible. We inquire to which extent this structural and computational simplicity extends beyond additive costs. Our main result is that the class of submodular cost functions admits an optimal strategy that follows a fixed order, thus preserving the structural simplicity of additive costs. In contrast, for the more general class of subadditive (or even fractionally subadditive) cost functions, the optimal strategy may inevitably determine the search order adaptively. On the computational side, obtaining any approximation to the optimal utility requires superpolynomially many queries to the cost function, even for a strict subclass of submodular cost functions. Funding: B. Berger and M. Feldman are partially supported by the HORIZON EUROPE European Research Council [Grant 866132] and the United States-Israel Binational Science Foundation [Grant 2020788]. T. Ezra is supported by the Harvard University Center of Mathematical Sciences and Applications. F. Fusco is partially supported by the National Recovery and Resilience Plan (PNRR) by Ministero dell’Università e della Ricerca [Project IR0000013-SoBigData.it] and the Future Artificial Intelligence Research project funded by the NextGenerationEU program within the National Recovery and Resilience Plan Extended Partnership for Artificial Intelligence (PNRR-PE-AI) scheme [Grant M4C2, investment 1.3, line on Artificial Intelligence]. Additionally, this work was supported by Ministero dell’Università e della Ricerca [Grant ALGADIMAR], the Israel Science Foundation [Grant 317/17], the Amazon Research Award, and the HORIZON EUROPE European Research Council [Grant 788893].

Bayesian Nonparametric Sequential Search

Sequential search models are popular in marketing for studying consumer search behavior. Current search models use parametric assumptions regarding different aspects of the models, such as random shocks, search costs, and consumer preferences. These assumptions can be restrictive. The authors develop a novel Bayesian nonparametric framework for sequential search to flexibly model unknown distributions. They also develop Markov chain Monte Carlo methods for inferring the model unknowns. The authors conduct simulation studies to demonstrate that currently popular parametric search models can yield incorrect estimates and inferences when the data-generating process deviates from their assumptions. In contrast, the proposed model accurately recovers these quantities for various data-generating processes. The methodology is then applied to online search and purchase data from a Japanese retailer. The results show that several heterogeneity distributions have complex patterns, such as multimodality and skewness, that are not captured by a parametric benchmark model. The authors also estimate the monetary value of search costs, the price elasticities, and a counterfactual profit gain under a personalized couponing strategy and find substantial differences in the results from the two models.

Searching for Manuscripts and Running a Business: Printing and the Search for Texts in the Fifteenth Century

This contribution seeks to examine how the commercial production of printed books interacted with the search for manuscripts of texts not otherwise known to survive, during the first fifty years after the European invention of printing. Searching for manuscripts is thus considered from the perspective of it being an economic activity. Producers of manuscript and of printed books alike had to locate and acquire a text to work from. This could range from the easy acquisition of a local exemplar, the acquisition of exemplars from known but distant locations, to the search for texts whose location was unknown, and all the way to the search for exemplars of texts whose survival was uncertain. By exploring the most ambitious types of search within this broader context, we will seek to understand better the circumstances under which the commercial production of printed books could enable a business model, one amongst many, that not only made such a search possible but even required it. We will seek to establish when the associated direct and indirect cost of an ambitious search could be a worthwhile investment, or at least could seem to be. In doing so we also aim to understand more clearly why this could be a potential path towards profit for commercial producers of printed books, while it would have been unviable for commercial producers of manuscript books.

AG-SpTRSV: An Automatic Framework to Optimize Sparse Triangular Solve on GPUs

Sparse Triangular Solve (SpTRSV) has long been an essential kernel in the field of scientific computing. Due to its low computational intensity and internal data dependencies, SpTRSV is hard to implement and optimize on graphics processing units (GPUs). Based on our experimental observations, existing implementations on GPUs fail to achieve the optimal performance due to their suboptimal parallelism setups and code implementations plus lack of consideration of the irregular data distribution. Moreover, their algorithm design lacks the adaptability to different input matrices, which may involve substantial manual efforts of algorithm redesigning and parameter tuning for performance consistency. In this work, we propose AG-SpTRSV, an automatic framework to optimize SpTRSV on GPUs, which provides high performance on various matrices while eliminating the costs of manual design. AG-SpTRSV abstracts the procedures of optimizing an SpTRSV kernel as a scheme and constructs a comprehensive optimization space based on it. By defining a unified code template and preparing code variants, AG-SpTRSV enables fine-grained dynamic parallelism and adaptive code optimizations to handle various tasks. Through computation graph transformation and multi-hierarchy heuristic scheduling, AG-SpTRSV generates schemes for task partitioning and mapping, which effectively address the issues of irregular data distribution and internal data dependencies. AG-SpTRSV searches for the best scheme to optimize the target kernel for the specific matrix. A learned lightweight performance model is also introduced to reduce search costs and provide an efficient end-to-end solution. Experimental results with SuiteSparse Matrix Collection on NVIDIA Tesla A100 and RTX 3080 Ti show that AG-SpTRSV outperforms state-of-the-art implementations with geometric average speedups of 2.12x ∼ 3.99x. With the performance model enabled, AG-SpTRSV can provide an efficient end-to-end solution, with preprocessing times ranging from 3.4 to 245 times of the execution time.

Lottery‐like features and mutual fund performance‐flow sensitivity

AbstractWe show that mutual funds' lottery‐like features weaken the performance‐flow sensitivity, particularly among low‐performing funds, thereby contributing to the convexity of the fund performance‐flow relation. The results hold when different model specifications are used to test the fund performance‐flow relation, are robust to alternative measures for funds' lottery‐like features, and cannot be attributed to fund search costs, marketing efforts, or fund performance volatility. Utilizing retail trading data at the account level from a large brokerage firm, we offer additional evidence that funds' lottery‐like features significantly reduce outflows for low‐performing funds. It confirms that the weakened performance‐flow sensitivity among low‐performing funds with lottery‐like features is driven by existing investors' reluctance to redeem their shares. Furthermore, we reveal that fund managers can cater to investors' gambling preference by tilting fund portfolios toward lottery‐type stocks. Funds' lottery‐like features, however, aggravate future fund performance, especially among those that have already underperformed in the past.

Central place foragers, prey depletion halos, and how behavioral niche partitioning promotes consumer coexistence

Many seabirds congregate in large colonies for breeding, a time when they are central place foragers. An influential idea in seabird ecology posits that competition during breeding results in an area of reduced prey availability around colonies, a phenomenon known as Ashmole's halo, and that this limits colony size. This idea has gained empirical support, including the finding that species coexisting within a colony might be able to do so by foraging on a single prey species but at different distances. Here, we provide a comprehensive mathematical model for central place foragers exploiting a single prey in a two-dimensional environment, where the prey distribution is the result of intrinsic birth and death, movement in space, and mortality due to foraging birds (we also consider a variant tailored toward colonial social insects). Bird predation at different distances occurs according to an ideal free foraging distribution that maximizes prey delivery under flight and search costs. We fully characterize the birds' ideal free distribution and the prey distribution it generates. Our results show that prey depletion halos around breeding colonies are a robust phenomenon and that the birds' ideal free distribution is sensitive to prey movement. Furthermore, coexistence of several seabird species on a single prey easily emerges through behavioral niche partitioning whenever trait differences between species entail trade-offs between efficiently exploiting a scarce prey close to the colony and a more abundant prey far away. Such behavioral-based coexistence-inducing mechanism should generalize to other habitat and diet choice scenarios.

Identity-based authenticated multi-keyword searchable encryption scheme for cloud-assisted telemedicine information system

In cloud-assisted telemedicine information system (CA-TMIS), patients upload data containing their health metrics to the cloud. Doctors can access the patient’s data through the cloud, thereby alleviating the geographical restrictions in the traditional medical system. To ensure the privacy of the patient, the patient will encrypt the data before uploading the data. How doctors can search through numerous encrypted files to locate specific patients’ relevant data is a topic worthy of research. Searchable encryption technology can fulfill the function of ciphertext retrieval. Some existing searchable encryption schemes use single-keyword for searching, resulting in insufficient search accuracy and increased search costs for doctors. Multi-keyword searchable encryption schemes can address this issue. Therefore, in this paper, we propose an identity-based authenticated multi-keyword searchable encryption scheme, which leverages multiple keywords for combined searching, enhancing search accuracy. The scheme is proven to be resistant to keyword guessing attacks in the standard model. Compared to four other existing schemes, our scheme does not utilize pairing operations and offers higher efficiency advantages. Consequently, our scheme is more suitable for CA-TMIS.

‘They adopt, I also adopt’: The neighborhood effects and irrigator farmers’ conversion to adopt water-saving irrigation technology

Promoting water-saving irrigation technology (WSIT) has long been recognized as an effective measure to reduce irrigation water use and alleviate regional water poverty. Neighbours are the groups with the most interaction, the most intensive contact and the most frequent communication in agricultural production. In this paper, based on a field survey of irrigator farmers of China, experiments were conducted to evaluate the contribution of neighbour effect to farmers’ WSIT behaviour and its moderating mechanism between neighbour effect and technology adoption. The results show 26.7 % farmers adopt water-saving technology and these farmers often have strong neighborly relationships. In particular, neighbor effect significantly promote WSIT adoption by 42.1 % and farmers with strong neighborly effects apply significant higher technology compared to farmers with weak neighbourly effects. Furthermore, mechanism results indicated that neighbour effects indirectly influence farmers' adoption of water-saving irrigation technology through three pathways: reducing information search costs, mitigating agricultural production risks and benefiting from the demonstration effect. Moreover, the heterogeneous results report that higher-income, lower digital and lower training frequency farmers will benefit more from the NE in terms of WSIT. Overall, this research provides a micro foundation and policy implications for the promotion of WSIT, and sheds light upon how the government can formulate relevant policies to promote the sustainable development of agricultural water resources.

The role of online search platforms in scientific diffusion

AbstractAfter the launch of Google Scholar older papers experienced an increase in their citations, a finding consistent with a reduction in search costs and introduction of ranking algorithms. I employ this observation to examine how recombination of science takes place in the era of online search platforms. The findings show that as papers become more discoverable, their knowledge is diffused beyond their own broad field. Results are mixed when examining knowledge diffusion within the same field. The results contribute to the ongoing debate of narrowing of science. While there might a general reduction in recombination of knowledge across distant fields over the last decades, online search platforms are not the culprits.

Verifiable privacy-preserving semantic retrieval scheme in the edge computing

Edge computing, with its characteristics of low latency and low transmission costs, addresses the storage and computation challenges arising from the surge in network edge traffic. It enables users to leverage nearby edge servers for data outsourcing and retrieval. However, data outsourcing poses risks to data privacy. Although searchable encryption is proposed to secure search of outsourced data, existing schemes generally cannot meet the requirements of semantic search, and they also exhibit security risks and incur high search costs. In addition, edge servers may engage in malicious activities such as data tampering or forgery. Therefore, we propose a verifiable privacy-preserving semantic retrieval scheme named VPSR suitable for edge computing environments. We utilize the Doc2Vec method to extract text feature vectors and then convert them into matrix form to reduce storage space requirements for indexes, queries, and keys. We encrypt matrices using an improved secure k-nearest neighbor (kNN) algorithm based on learning with errors (LWE) and calculate text similarity by solving the Hadamard product between matrices. Additionally, we design an aggregable signature scheme and offload part of the result verification tasks to edge servers. Security and performance analysis results demonstrate that the VPSR scheme is suitable for edge computing environments with high encryption and search efficiency and low storage cost while ensuring security.

Does distance from government hinder enterprises’ OFDI？Evidence from China

This study analyzes the impact of government–enterprise distance on the probability of outward foreign direct investment (OFDI) with matched data from Chinese prefecture-level municipal governments, the Chinese industrial enterprises database, and the directory of overseas investment enterprises. We determine that increased distance between the government and enterprises has a significant negative impact on the probability of enterprises’ OFDI. Mechanism analysis demonstrates that increased government–enterprise distance, which causes higher information search and relationship building costs, and decreased accessibility of public service, are the main reasons for this decline.

A Study on Information Search Behavior Using AI-Powered Engines: Evidence From Chatbots on Online Shopping Platforms

The development of Artificial Intelligence (AI) has significantly influenced how consumers search for information. However, there is a lack of comprehensive models based on theoretical foundations that specifically address AI-powered information search behavior. This study draws on psychological motivation, information processing, and information economics theories to develop a theoretical model of consumer AI-powered information search behavior. The study aims to identify the main factors affecting consumer search behavior, offering a more holistic understanding of consumer behavior in the context of AI. Analyzing 512 valid questionnaires, the study shows that search motivation not only had the most significant impact on search behavior but also served as a mediator between other variables and search intensity. Additionally, perceived search ability had a direct and the greatest indirect impact on search behavior, while other variables such as perceived search costs and benefits also had indirect effects on search behavior. Practically, the study offers valuable insights for businesses and AI developers. Understanding the factors that drive AI-powered search behavior can inform the design of more effective AI systems and marketing strategies.

The contributions of direct and indirect selection to the evolution of mating preferences.

Many influential mathematical models of sexual selection have stressed that mating preferences evolve due to correlations that build between mating preferences and preferred display traits-that is, through indirect selection. Nevertheless, there is a perception that indirect selection should generally be overwhelmed by direct selection, for example, in the form of search costs. Recent work by Fry has used quantitative genetic models to argue that in many cases, including when there are direct benefits (a fecundity advantage to mating with the preferred male), direct and indirect selection may be of similar magnitude. Here, I use population genetic models, in which the strength of the genetic correlation is an emergent property of evolution at mating preference and display trait loci, to assess the relative contributions of direct and indirect selection to the evolution of mating preferences. For the cases of direct benefits and of indirect benefits with fixed and frequency-dependent search costs, I outline parameter values of fecundity benefits, preference strengths, and search costs for which indirect selection on female preferences can potentially predominate. I also analyze male mate choice under polygyny, showing that direct selection will always outweigh indirect selection except when there are direct benefits.

The impact of health and environmental messaging with and without product filtering in complex retail markets: the case of pulses.

Multiple barriers exist to healthy and environmentally sustainable food choices. Limited consumer understanding of the health and environmental implications of food choices complicates their abilities to make choices that lead to desired outcomes. The complexity of the retail environment itself may crowd out less immediate motivations to address health or environment. Even if consumers understand general impacts of food choices on health and environmental outcomes, there may be non-negligible time and search costs to identifying the products that meet consumers' needs. In many food categories, the foods containing attributes that help achieve health and sustainability outcomes may represent a small percentage of available products. In this research, we examine the case of pulses-beans, chickpeas, lentils, and dry peas. Pulses are nutritious and have a low environmental impact. However, consumption of pulses in the US is quite low, which may be attributable to low consumer knowledge of pulse benefits, as well as difficulty of identifying pulse products in retail environments. In this research, we examine the choice of pulse-based foods in three conditions: (1) a control condition, (2) a messaging condition communicating the health and environmental benefits of pulse products, and (3) a paired messaging condition with a choice environment intervention that allows respondents to choose to filter products to those that contain pulses. Participants selected a food item from each of six food categories. We find slight, but significant, increases in pulse choice in the messaging only condition relative to the control condition, but dramatic and highly significant increases when participants can filter the products to easily view pulse products. We also find evidence for knowledge being a barrier to healthy/sustainable food choice. Participants exposed to the messaging were more likely to view pulses as environmentally beneficial, and less likely to report that they did not know the health or environmental impacts of pulse foods. We find that paired messaging and filtering interventions significantly increase the choice of pulse-based foods, which offer both human health and sustainability benefits.