Financial and knowledge barriers to innovation: Complementary and substitution effects on innovative effort     

《Research Policy》2023,52(7):104814

Innovation studies have thoroughly analysed how firms' perception of barriers affects their innovative behaviour. We contribute to this line of research by analysing complementary and substitution effects between financial barriers, internal knowledge barriers and external knowledge barriers on firms' innovative effort. Using a panel data set of Uruguayan firms between 2006 and 2015, we add to empirical studies on this topic by estimating both unconditional and conditional complementary and substitution effects between these innovation barriers. We apply an econometric two-step specification that includes an estimation of instrumental variables, which allows us to control for endogeneity biases in the supermodularity and submodularity tests. We find significant evidence that external and internal knowledge barriers reinforce each other's effects (complementary effects), conditional on the absence of financial barriers. We also find that financial and internal knowledge barriers mitigate each other's effects (substitution effects) conditional on the firm's perception of external knowledge barriers. Additionally, we find substitution effects between financial and external knowledge barriers, conditional on the firm's perception of internal knowledge barriers. We discuss these results against the backdrop of Uruguayan innovation policies during the period studied, and stress the necessary articulation between instruments that will help strengthen firms' innovation management to recognize and manage innovation barriers.  相似文献   

Affiliation bias in peer review and the gender gap     

《Research Policy》2023,52(7):104797

We document an affiliation bias in top Chinese journals and the gender gap in the current affiliation effect. We find that papers written by researchers who are affiliated with the journals' host institutions on average receive fewer citation counts than those written by non-affiliated researchers. Moreover, we show that this affiliation bias is greater for men than for women. We propose that the current effect can be accounted for by a social capital mechanism, and we provide evidence consistent with our proposition. In addition, the results of our analyses demonstrate heterogeneity of the affiliation bias and the gender gap with regard to researcher status. We discuss alternative mechanisms of the current effect and implications of our research.  相似文献   

Private and social functions of patents: Innovation,markets, and new firms     

《Research Policy》2023,52(7):104806

This article provides a review of the private and social functions of patents using data and evidence from the economic and management literature. While patents provide incentives to invent by providing private protection to appropriate the returns on inventions, they also have broader effects. For example, in this paper we focus on the fact that they provide signals about the value of new firms, disclose information about the invention, and encourage the exchange of inventions and ideas in markets for technology. In order to better understand the relative importance of the implications of patents, patent agencies and stakeholders should invest to a greater extent in data collections or in creating the conditions for research designs and experiments that nail down causal effects and mechanisms. Available data are not created with these identification strategies in mind, which limits the questions that scholars can ask. Systematic studies that identify different effects of patents can provide the basis for rigorous evidence-based management and policy about patents. This would imply a wider shift from a world in which managerial and policy analysis is distinct from practice, to a world in which analysis and implementation are increasingly co-produced, and there is greater integration between them.  相似文献   

Performance-based research funding: Evidence from the largest natural experiment worldwide     

《Research Policy》2023,52(6):104780

A performance-based research funding system (PRFS) is a nationwide incentive scheme that promotes and rewards university research performance through competition for government funding. The UK’s PRFS, currently the Research Excellence Framework (REF), is considered the oldest, largest and most developed payment-by-results system in academia worldwide. Surprisingly, and despite the strong criticisms, little has been done to quantitatively and casually evaluate the intended and unintended effects of the PRFSs. In this paper, we evaluate the incremental impact of the REF 2014 in the fields of Economics and Business. We use a synthetic control method to compare the performance of UK universities with their artificial counterfactual units constructed using data from US universities. Our analysis shows, on the whole, that the introduction of the REF had a significant and positive impact on the quantity and quality of the scientific research produced at UK universities. However, we do not find a significant effect on the per author measures, suggesting that the REF did not result in an increase in research productivity. We also show that the effects are more heterogeneous across universities than across academic disciplines. We do not find evidence of a shift of research focus from Economics to Business topics, as some feared. But our analysis indicates that the REF 2014 may have contributed to the concentration of research excellence in elite institutions.  相似文献   

Big data and dynamic capabilities in the digital revolution: The hidden role of source variety     

《Research Policy》2023,52(7):104812

Recent research frames big data as a resource enhancing dynamic capabilities through improved prediction, decision-making, and data-driven innovation. In contrast, this study frames big data as an evolutionary driver that channels firms' knowledge and attention in specific directions, implying that firms need multiple big data sources to be receptive and dynamically capable. I apply this framework to the context of the digital revolution and focus on the impact of big data on firms' digitalization priorities. By leveraging a large-scale survey of more than twenty thousand Italian firms of all sizes, I find that big data improves the digitalization awareness of firms only if they gather big data from more than one source (otherwise, counterintuitively, it may even decrease it). I also find a positive effect of source variety both on the likelihood of prioritizing individual digitalization factors and on the variety of digitalization factors prioritized. Such effects appear to be stronger for small firms relative to their larger counterparts. Given the path dependence of digitalization trajectories, these findings have relevant policy implications in the context of initiatives like the European strategy for data and the SME strategy for a sustainable and digital Europe.  相似文献   

Zero-shot cross-lingual transfer language selection using linguistic similarity     

《Information processing & management》2023,60(3):103250

We study the selection of transfer languages for different Natural Language Processing tasks, specifically sentiment analysis, named entity recognition and dependency parsing. In order to select an optimal transfer language, we propose to utilize different linguistic similarity metrics to measure the distance between languages and make the choice of transfer language based on this information instead of relying on intuition. We demonstrate that linguistic similarity correlates with cross-lingual transfer performance for all of the proposed tasks. We also show that there is a statistically significant difference in choosing the optimal language as the transfer source instead of English. This allows us to select a more suitable transfer language which can be used to better leverage knowledge from high-resource languages in order to improve the performance of language applications lacking data. For the study, we used datasets from eight different languages from three language families.  相似文献   

Understanding AI-based customer service resistance: A perspective of defective AI features and tri-dimensional distrusting beliefs     

《Information processing & management》2023,60(3):103257

Communicating with customers through AI-based chatbots in customer service (AISC) has become increasingly popular for many companies. However, in actual service encounters, AISC seems defective and is not always accepted by customers. Occasionally it is even resisted. This study aims to investigate such customer resistance. In addition to two cognition-centered AI features (i.e., irrelevant and biased information) discussed in prior studies, this study proposes that lack of empathy is another key feature of defective AI (i.e., in its emotional dimension) and investigates the underlying mechanism of empathy. Specifically, this study proposes three pathways in which empathy functions are lacking. A survey was conducted to test our hypotheses, and the results suggest that lack of empathy has three effects on customer resistance: direct, indirect, and moderating. Finally, theoretical contributions and practical implications are discussed.  相似文献   

上海交通大学数字图书馆雏型   总被引：14，自引：2，他引：14  

杨宗英  叶爱芬  王榕榕  郑巧英 《现代图书情报技术》1998,14(5):3-6

简要介绍上海交通大学数字图书馆雏型产生背景和概念模型, 它包括五个模块: 各种载体数字化, 大容量数据储存和管理, 数据访问, 数字化资料传递, 权限管理和版权保护; 上海交通大学数字图书馆雏型的体系结构包括网络拓扑结构、主干网的缚线系统和硬软件环境等; SJTUDL 的实用服务模式由数字化图书馆的特种馆藏, 商用网上联机电子出版物和Internet 文献信息资源三部分组成, 将用一个统一的界面向读者提供服务。  相似文献   

研究生与导师冲突的类型特征及其化解——基于20个案例的分析     

曾剑雄  张国栋 《复旦教育论坛》2023,21(1):36-43

基于2016年以来20起研究生与导师的冲突事件,结合社会冲突理论,本研究将当前导生冲突划分为三种类型特征,即权利冲突、利益冲突以及道德冲突。权利使用与责任履行的非对等、充分培养与过度使用的不平衡、维权意识与化解渠道的不匹配是造成导生冲突的重要因素。化解导生冲突的路径有：加强师德师风建设,营造平等和谐的良好氛围;明晰导生权责边界,压缩导生冲突产生的行为空间;构建制度化的冲突解决机制,引导研究生理性维权。  相似文献   

一种用于电类实验测量数据判决的改进KNN算法北大核心     

申赞伟  周军盈  张士文  殳国华  张峰 《实验技术与管理》2022,(1):18-23

电类实验教学过程中人工评判学生所测数据工作烦琐,影响了教学质量和效率。该文提出了改进的K近邻(K-nearest neighbors,KNN)分类算法,即基于均值漂移、安全间隔和核主成分分析(KPCA)的M-KPCA-KNN(KNN based on margin and KPCA)算法,以判断学生测量数据正确与否和错误原因。首先利用KPCA对高维实验数据进行降维,然后利用均值漂移向量找到不同类别数据的最密集位置,并在不同类别数据的边界设置安全间隔,最后,将与待测样本距离最近的k个数据设置权重,计算每个类别的权重和,权重和最大的类别为待测样本的类别。与现有的KNN算法相比,M-KPCA-KNN算法不仅提高了分类正确率,而且降低了时间复杂度。  相似文献