The first six articles in this issue of Transactions in GIS were gathered from a call for abstracts and will be presented in two research sessions scheduled during the 2021 Esri User Conference. From the 19 abstracts submitted, the journal editors selected six for preparation as full journal articles. Each of these manuscripts passed through the usual journal peer review process and the final versions included here have been revised in light of both the reviewers' and the editors' feedback.

The six articles included in this issue cover a wide range of topics and address some of the fundamental concepts and applications of geographic information science from a variety of perspectives. The first offers a review that the authors hope will expand and solidify the scope of place-based GIS. The second proposes a privacy-preserving framework for location recommendation using decentralized collaborative machine learning. The third proposes a method to extract rules for agent-based models from movement data using classification trees and build data-driven simulations of the movements of an olive baboon troop. The fourth uses GIS to reconstruct spatiotemporal events from narratives and decide whether scenarios are conceivable (or not) in the narrative world. The fifth uses GIS and convolutional neural networks to extract buildings in support of humanitarian work. The sixth proposes a new soft computing logic method and tool for site suitability analysis.

In the first article, by Vicente Tang and Marco Painho, the authors present a literature review to describe the distinct but overlapping frameworks that scholars have proposed to bridge the gap between space, place and GIS. The review shows that most studies focus on knowledge-based models in urban settings drawing on concepts from human geography. This article synthesizes the current state-of-the-art to encourage new conceptual and methodological approaches to expand and solidify the scope of place-based GIS.

The second article, by Jinmeng Rao, Song Gao, Mingxiao Li, and Qunying Huang, presents a privacy-preserving framework for location recommendation using decentralized collaborative machine learning. This framework uses information about transportation infrastructure, public safety, and flow-based spatial interaction (similar to traditional centralized learning approaches) but keeps users' data on their own devices and trains the models locally. The framework then aggregates and updates local model parameters via secure multi-party computation to garner the benefits of collaborative learning among users while preserving privacy.

The third article, by Jugal Patel, Jeffrey Katan, Liliana Perez, and Raja Sengupta, leverages machine learning to determine the environmental features associated with moving behaviors of a troop of olive baboons in Kenya. Their workflow performs path segmentation using thresholding to label training data, an agent-based rule extraction using classification trees to associate the relative Euclidian distance between a point and environmental features with behavior, and implements this information in an agent-based model that provides data-driven simulations of troop movements toward set destinations. The authors conclude by noting that their framework is scalable and that it will be able to support larger and more varied data inputs in future applications.

The fourth article, by Vincent van Altena, Jan Krans, Henk Bakker, and Jantien Stoter, examines the use of GIS to reconstruct spatiotemporal events from narratives to examine whether a scenario is conceivable within the narrative world. The authors use least-cost path analysis, network analysis, and space-time cubes to examine the narrative about Paul’s escape from Berea (Acts 17:14–15) and they conclude that the aforementioned methods can help a modern reader to better understand and appreciate the conceivability of stories from the narrative world.

The fifth article, by Dirk Tiede, Gina Schwendemann, Ahmad Alobaidi, Lorenz Wendt, and Stefan Lang, uses a Mask R-CNN deep learning approach and Pléiades very high-resolution satellite imagery to extract 1.2 million dwellings and buildings for Khartoum, Sudan. Their method strikes a balance between the need for timely information during the COVID-19 pandemic and the accuracy of the result by combining the outputs of three different models tailored to specific types of buildings. This work, which aimed to support humanitarian organizations in response to the COVID-19 pandemic, illustrates the potential for using convolutional neural networks with GIS for dwelling extraction from satellite imagery.

The sixth and final article, by Shuoge Shen, Suzana Dragićević, and Jozo Dujmović, extends the logic scoring method of preference (LSP) as a general multicriteria evaluation method implemented within a GIS environment. The authors used an urban densification suitability analysis in the Metro Vancouver Region, Canada, to illustrate and validate the new LSP.GIS method. The results, in turn, show how this method provides a flexible and sensitive workflow for generating outcomes bounded by stakeholders’ goals and requirements.

These six articles, taken as a whole, illustrate the breadth and depth of geographic information science scholarship and best practice across a variety of settings, including a call to expand place-based GIS, new workflows to reconstruct spatiotemporal events from narratives and extract buildings from satellite data, and new methods to make privacy-preserving location recommendations, delineate the environmental features associated with wildlife mobility, and assess site suitability.

Special thanks are owed to the authors, and especially to those who provided the peer reviews, for helping to move the initial abstracts to full, peer-reviewed articles in just a few months. I trust that all involved will see how these contributions bore fruit when you read the final versions of the articles in this thirtieth issue of Transactions in GIS organized around several research sessions hosted by Esri and given a prominent place in its User Conference program.

本期GIS 交易中的前六篇文章是从征集摘要中收集的，并将在 2021 年 Esri 用户大会期间安排的两次研究会议上发表。从提交的 19 篇摘要中，期刊编辑选择了 6 篇作为完整的期刊文章进行准备。这些手稿中的每一个都通过了通常的期刊同行评审过程，这里包含的最终版本已经根据评审员和编辑的反馈进行了修订。

本期包含的六篇文章涵盖了广泛的主题，并从多个角度阐述了地理信息科学的一些基本概念和应用。第一个提供了评论，作者希望将扩大和巩固基于地点的 GIS 的范围。第二个提出了使用分散协作机器学习的位置推荐隐私保护框架。第三个提出了一种方法，使用分类树从运动数据中提取基于代理的模型规则，并构建橄榄狒狒部队运动的数据驱动模拟。第四个使用 GIS 从叙事中重建时空事件，并决定在叙事世界中情景是否可想象（或不可想象）。第五个使用 GIS 和卷积神经网络来提取建筑物以支持人道主义工作。第六个提出了一种新的软计算逻辑方法和工具，用于场地适宜性分析。

在 Vicente Tang 和 Marco Painho 撰写的第一篇文章中，作者提供了一篇文献综述来描述学者们提出的不同但重叠的框架，以弥合空间、地点和 GIS 之间的差距。审查表明，大多数研究都侧重于城市环境中基于知识的模型，这些模型借鉴了人文地理学的概念。本文综合了当前最先进的技术，以鼓励采用新的概念和方法论方法来扩展和巩固基于地点的 GIS 的范围。

由 Jinmeng Rao、Song Gao、Mingxiao Li 和 Qunying Huang 撰写的第二篇文章提出了一种使用去中心化协作机器学习进行位置推荐的隐私保护框架。该框架使用有关交通基础设施、公共安全和基于流的空间交互（类似于传统的集中式学习方法）的信息，但将用户的数据保存在自己的设备上并在本地训练模型。然后，该框架通过安全的多方计算聚合和更新本地模型参数，以在保护隐私的同时获得用户之间协作学习的好处。

第三篇文章由 Jugal Patel、Jeffrey Katan、Liliana Perez 和 Raja Sengupta 撰写，利用机器学习来确定与肯尼亚橄榄狒狒队伍的移动行为相关的环境特征。他们的工作流程使用阈值来标记训练数据执行路径分割，基于代理的规则提取使用分类树将点和环境特征之间的相对欧几里德距离与行为相关联，并在基于代理的模型中实现这些信息，提供数据-驱动模拟部队向既定目的地移动。作者最后指出，他们的框架是可扩展的，并且能够在未来的应用程序中支持更大、更多样化的数据输入。

由 Vincent van Altena、Jan Krans、Henk Bakker 和 Jantien Stoter 撰写的第四篇文章探讨了如何使用 GIS 从叙事中重建时空事件，以检验在叙事世界中是否可以想象某个场景。作者使用最低成本路径分析、网络分析和时空立方体来检查关于保罗逃离庇哩亚的叙述（使徒行传 17:14-15），他们得出结论，上述方法可以帮助现代读者更好地理解和欣赏叙事世界中故事的可想象性。

由 Dirk Tiede、Gina Schwendemann、Ahmad Alobaidi、Lorenz Wendt 和 Stefan Lang 撰写的第五篇文章使用 Mask R-CNN 深度学习方法和Pléiades超高分辨率卫星图像为苏丹喀土穆提取了 120 万套住宅和建筑物。他们的方法通过结合针对特定类型建筑物量身定制的三种不同模型的输出，在 COVID-19 大流行期间对及时信息的需求与结果的准确性之间取得了平衡。这项旨在支持人道主义组织应对 COVID-19 大流行的工作说明了使用带有 GIS 的卷积神经网络从卫星图像中提取住宅的潜力。

由 Shuoge Shen、Suzana Dragićević 和 Jozo Dujmović 撰写的第六篇也是最后一篇文章将偏好的逻辑评分方法 (LSP) 扩展为在 GIS 环境中实施的通用多标准评估方法。作者使用加拿大大温哥华地区的城市密集化适宜性分析来说明和验证新的 LSP.GIS 方法。结果反过来表明，该方法如何提供灵活且敏感的工作流程，以生成受利益相关者目标和要求限制的结果。

这六篇文章作为一个整体，说明了地理信息科学学术研究和各种环境中最佳实践的广度和深度，包括扩展基于地点的 GIS、从叙述中重建时空事件的新工作流程以及从卫星数据，以及提出保护隐私的位置建议的新方法，描绘与野生动物移动相关的环境特征，并评估站点的适宜性。

特别感谢作者，尤其是那些提供同行评审的人，他们帮助在短短几个月内将最初的摘要变成了完整的、经过同行评审的文章。我相信，当您阅读第 30 期GIS 交易中的文章的最终版本时，所有相关人员都会看到这些贡献如何取得成果，这些文章围绕 Esri 主办的几个研究会议组织，并在其用户会议计划中占据突出位置。