基于空间网络的轨迹相似性搜索研究

With the rapid development of GPS-enabled mobile devices and online map services, trajectory data is pervasive in our daily life. Trajectory similarity search is a significant research problem in trajectory data management. This type of search is useful in many popular mobile applications, such as carpooling, trip planning and recommendation, etc., and can benefit many areas, including smart city, intelligent transportation system, and location based services in general. This type of search is making travel more convenient and is helpful in improving transportation resource allocation, controlling traffic jams, reducing resource consumption, and reducing air pollution. ..However, most existing studies only focus on Euclidean space, and their search results can hardly be used in real application scenarios directly (since most application scenarios are network-based, e.g., in road networks). In addition, user requirements are becoming more and more specific, and how to integrate user preferences with trajectory similarity models is the main challenge in our research. To overcome these challenges, we propose a novel project named "trajectory similarity search in spatial networks" to investigate the key technologies in trajectory similarity search under massive data environment. First, we will propose a novel trajectory similarity model, which can be used in spatial networks and can reflect user preferences. Second, we will design and implement I/O efficient algorithms based on spatial database technologies to support the trajectory similarity search in spatial networks. Third, we will devise a further optimization strategy to improve the unsatisfied search results. Finally, we will integrate the proposed models and algorithms with moving object databases, and will conduct extensive experiments to verify the performance of trajectory similarity search.

随着全球定位技术和在线地图服务的高速发展，轨迹数据在我们的生活中无处不在。轨迹相似性搜索是轨迹数据管理中的关键性问题，此类搜索可广泛应用在拼车出行、行程规划与推荐等服务，乃至智慧城市、智能交通系统、基于位置的服务等领域，有助于方便个人出行、优化交通资源配置、缓解交通拥堵和节能减排。已有的轨迹搜索研究主要集中在欧式空间，其技术手段不适用于空间网络环境，因此很难被大多数实际应用所采纳。随着用户需求的不断细化，如何在轨迹搜索中体现用户个人偏好是当前研究所面临的又一个挑战。本项目针对上述挑战，面向海量轨迹数据，研究基于空间网络的轨迹相似性搜索中的关键技术。首先提出能适应空间网络特点、并能体现用户偏好的轨迹相似度模型；然后利用空间数据库技术，设计并实现多维度环境下的高性能轨迹搜索算法以突破I/O瓶颈；之后提出针对不理想搜索结果的整合优化策略；最后在理论方法研究的基础上，与移动对象数据库进行集成验证。

随着全球定位技术和在线地图服务的高速发展，轨迹数据在我们的生活中无处不在。轨迹相似性搜索是轨迹数据管理中的关键性问题，此类搜索可广泛应用在拼车出行、行程规划与推荐等服务，乃至智慧城市、智能交通系统、基于位置的服务等领域，有助于方便个人出行、优化交通资源配置、缓解交通拥堵和节能减排。已有的轨迹搜索研究主要集中在欧式空间，其技术手段不适用于空间网络环境，因此很难被大多数实际应用所采纳。随着用户需求的不断细化，如何在轨迹搜索和路径规划中体现用户个人偏好是当前研究所面临的又一个挑战。本项目针对上述挑战，面向海量时空数据，研究基于空间网络的轨迹搜索和路径规划中的关键技术，提出并实现高性能的搜索算法，并将算法与系统有效集成并验证搜索效率。本项目在国际重要期刊和会议上共发表SCI、EI检索论文11篇。