Abstract
ABSTRACT In conventional cartography, while base maps are directly produced, smaller scale maps are derived through cartographic generalization. Efforts to automate the process of generalization began in the 1970s; however, it faced challenges due to its uncertainty nature. The generalization commences with the object selection/elimination process. The goal of this study is to once again bring to light the inherent uncertainty present in the selection/elimination process. To this end, the streams selected by two algorithms and seven cartographers in addition to the streams in the original 50K were compared according to geometric properties, performance metrics, and quantitative geomorphic measurements. Our findings emphasize the varying nature of outcomes: no two results were completely identical, underscoring the variable nature of manual and digital generalization. This variation reveals the need for balancing human intuition with algorithmic consistency in the selection of geographical objects, a challenge that has broad implications for the future of cartographic generalization.