GeoVectors

GeoVectors是由L3S研究中心和汉诺威莱布尼茨大学创建的一个全球规模的开放链接语料库，涵盖了OpenStreetMap中超过9.8亿个地理实体的嵌入表示，覆盖180个国家。该数据集通过捕捉地理实体的语义和地理维度，为机器学习算法和语义应用提供了直接访问这些实体的能力。创建过程中，利用了现有的表示学习技术，如词嵌入和地理表示学习，以捕捉OpenStreetMap数据的多个方面。GeoVectors的应用领域广泛，旨在解决地理信息检索、机器学习应用和知识图谱嵌入等问题，通过提供预计算的地理实体表示，极大地简化了这些领域中对OpenStreetMap实体的使用。

GeoVectors is a large-scale open linked corpus developed by the L3S Research Center and Leibniz University Hannover. It contains embedding representations of over 980 million geographic entities from OpenStreetMap, covering 180 countries worldwide. This dataset enables direct access to these entities for machine learning algorithms and semantic applications by capturing both the semantic and geographic dimensions of geographic entities. During its development, existing representation learning techniques, such as word embeddings and geographic representation learning, were employed to capture multiple aspects of OpenStreetMap data. GeoVectors has a wide range of application scenarios, aiming to address issues in geographic information retrieval, machine learning applications, and knowledge graph embedding. By providing pre-computed geographic entity representations, it greatly simplifies the utilization of OpenStreetMap entities in these fields.