neolifer/TripWorld

Trip World是一个公开可重分发的、METRO级别的跨城市旅行基准数据集。它结合了大量异地旅行序列、丰富的Foursquare和Google Maps兴趣点（POI）元数据以及统一的Google评论语料库。该数据集旨在支持：轨迹推荐（在陌生城市推荐POI序列，基于旅行者的家乡城市行为和期望的开始/结束/长度）、跨城市POI排名和下一个POI预测（根据异地游客的可能性对目的地区域的POI进行评分）、地点建模（基于1.86亿条Google评论以及68.7万个POI的结构化FSQ和Google属性进行文本/类别/情感分析）、旅游和移动性分析（对90多个国家1,173个城市之间的旅行流量进行聚合分析）。数据集基于STD-2018签到流构建，并应用了地铁整合、家乡发现、旅行者过滤和POI增强。Trip World is a publicly redistributable, METRO-level cross-city travel benchmark dataset. It integrates a large number of cross-city travel sequences, rich metadata of Points of Interest (POIs) from Foursquare and Google Maps, and a unified Google review corpus. This dataset is designed to support the following tasks: trajectory recommendation (recommending POI sequences in unfamiliar cities based on a traveler's home-city behavioral patterns and desired start/end points as well as trip length), cross-city POI ranking and next-POI prediction (scoring POIs in target regions based on the likelihood of attracting out-of-town visitors), place modeling (conducting text, category, and sentiment analysis based on 186 million Google reviews and structured Foursquare (FSQ) and Google attributes of 687,000 POIs), and tourism and mobility analysis (performing aggregated analysis of travel flows across 1,173 cities in over 90 countries). The dataset is constructed based on the STD-2018 check-in stream, with applications of metro integration, home location discovery, traveler filtering, and POI enhancement.