Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher.
Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?
Some links on this page may take you to non-federal websites. Their policies may differ from this site.
-
Free, publicly-accessible full text available December 1, 2025
-
Free, publicly-accessible full text available November 12, 2025
-
Free, publicly-accessible full text available November 12, 2025
-
Free, publicly-accessible full text available September 25, 2025
-
Free, publicly-accessible full text available August 24, 2025
-
Free, publicly-accessible full text available December 31, 2025
-
Free, publicly-accessible full text available August 10, 2025
-
Personalized recommender systems play a crucial role in modern society, especially in e-commerce, news, and ads areas. Correctly evaluating and comparing candidate recommendation models is as essential as constructing ones. The common offline evaluation strategy is holding out some user-interacted items from training data and evaluating the performance of recommendation models based on how many items they can retrieve. Specifically, for any hold-out item or so-called target item for a user, the recommendation models try to predict the probability that the user would interact with the item and rank it among overall items, which is called
global evaluation . Intuitively, a good recommendation model would assign high probabilities to such hold-out/target items. Based on the specific ranks, some metrics likeRecall@K andNDCG@K can be calculated to further quantify the quality of the recommender model. Instead of ranking the target items among all items, Koren first proposed to rank them among a smallsampled set of items , then quantified the performance of the models, which is calledsampling evaluation . Ever since then, there has been a large amount of work adopting sampling evaluation due to its efficiency and frugality. In recent work, Rendle and Krichene argued that the sampling evaluation is “inconsistent” with respect to a global evaluation in terms of offline top-K metrics.In this work, we first investigate the “inconsistent” phenomenon by taking a glance at the connections between sampling evaluation and global evaluation. We reveal the approximately linear relationship between sampling with respect to its global counterpart in terms of the top-
K Recall metric. Second, we propose a new statistical perspective of the sampling evaluation—to estimate the global rank distribution of the entire population. After the estimated rank distribution is obtained, the approximation of the global metric can be further derived. Third, we extend the work of Krichene and Rendle, directly optimizing the error with ground truth, providing not only a comprehensive empirical study but also a rigorous theoretical understanding of the proposed metric estimators. To address the “blind spot” issue, where accurately estimating metrics for small top-K values in sampling evaluation is challenging, we propose a novel adaptive sampling method that generalizes the expectation-maximization algorithm to this setting. Last but not least, we also study the user sampling evaluation effect. This series of works outlines a clear roadmap for sampling evaluation and establishes a foundational theoretical framework. Extensive empirical studies validate the reliability of the sampling methods presented.Free, publicly-accessible full text available March 31, 2025 -
Free, publicly-accessible full text available May 13, 2025
-
Abstract Population change is a main driver behind global environmental change, including urban land expansion. In future scenario modeling, assumptions regarding how populations will change locally, despite identical global constraints of Shared Socioeconomic Pathways (SSPs), can have dramatic effects on subsequent regional urbanization. Using a spatial modeling experiment at high resolution (1 km), this study compared how two alternative US population projections, varying in the spatially explicit nature of demographic patterns and migration, affect urban land dynamics simulated by the Spatially Explicit, Long-term, Empirical City development (SELECT) model for SSP2, SSP3, and SSP5. The population projections included: (1) newer downscaled state-specific population (SP) projections inclusive of updated international and domestic migration estimates, and (2) prevailing downscaled national-level projections (NP) agnostic to localized demographic processes. Our work shows that alternative population inputs, even those under the same SSP, can lead to dramatic and complex differences in urban land outcomes. Under the SP projection, urbanization displays more of an extensification pattern compared to the NP projection. This suggests that recent demographic information supports more extreme urban extensification and land pressures on existing rural areas in the US than previously anticipated. Urban land outcomes to population inputs were spatially variable where areas in close spatial proximity showed divergent patterns, reflective of the spatially complex urbanization processes that can be accommodated in SELECT. Although different population projections and assumptions led to divergent outcomes, urban land development is not a linear product of population change but the result of complex relationships between population, dynamic urbanization processes, stages of urban development maturity, and feedback mechanisms. These findings highlight the importance of accounting for spatial variations in the population projections, but also urbanization process to accurately project long-term urban land patterns.
Free, publicly-accessible full text available March 15, 2025