A media-informed global dataset on conflict/cooperation dynamics in transboundary rivers

Cooperation is fundamental to the functioning of human societies. Various cooperation models have been constructed with the tools of game theory to model communication (negotiation) processes among actors, generating conflict/cooperation behaviour. However, due to lack of large-scale empirical datasets, rigorous empirical tests of these models are rare, limiting their usefulness for managing social conflicts. We developed a global dataset on conflict/cooperation communication events in transboundary river management including six most salient elements: involved actors (“who” and “to whom”), discussed issues (“say what”), interaction modes (“through which channel”), sentiment polarities of conflict/ cooperation (“with what effects”), the basin context (“in which basin”) and when. This dataset contains 9382 communication events obtained by human coding from the newspaper articles included in over 760 media outlets across 105 transboundary rivers involving 84 countries or regions, with a time coverage from 1977 to 2022. This dataset provides the first-hand data for development, calibration and validation of cooperation communication process models for improving the modelling capacity, thus assist in mitigating the conflicts facing the world.

合作是人类社会运转的根本基石。学界已借助博弈论工具构建了多种合作模型，用以刻画行为体间的沟通（谈判）过程，进而推演冲突与合作行为。然而，由于缺乏大规模实证数据集，针对这些模型的严谨实证检验极为稀缺，这限制了其在社会冲突治理中的应用价值。我们构建了一套面向跨境河流管理领域的冲突/合作沟通事件全球数据集，涵盖六大核心要素：涉事行为体（"谁"与"向谁"）、讨论议题（"谈论内容"）、互动模式（"通过何种渠道"）、冲突/合作的情感极性（"产生何种影响"）、流域背景（"所属流域"）以及事件发生时间。该数据集包含9382条沟通事件，均由人工对覆盖105条跨境河流、84个国家或地区的760余家全球媒体机构刊发的新闻报道进行编码标注所得，时间跨度为1977年至2022年。本数据集为合作沟通过程模型的开发、校准与验证提供了一手数据，有助于提升相关建模能力，进而助力缓解全球面临的各类社会冲突问题。