基于位置服务的多维度空间信息获取与检索

As an emerging technology, mobile search has attracted significant attention from both the academic and industrial community, thanks to the modern smart phones and the Internet. Location-based service is an irreplaceable functionality in mobile search. According to statistics, in December 2011, 95% smart-phone users have used location-based service and more than 50% location-based search is from mobile users. Recently, location-based service has been widely accepted by mobile users. However existing location-based search methods have some limitations and disadvantages. To address these problems, in this proposal, we propose multi-dimensional data extraction and search to improve location-based services and enhance user experiences. This proposal mainly studies the following problems. First, as traditional methods only use limited unstructured data (e.g., Point of Interests (POIs)) and cannot provide much richer structured information (e.g., price, rating, and telephone numbers for restaurants). In this proposal, we study how to extract high-quality multi-dimensional structured data for POIs from the web in order to provide users with detailed information. Second, we propose to deeply integrate the location information and textual description based on the multi-dimensional data and utilize semantics of the queries and data to improve the search quality. Third, we extract social information for mobile users and utilize the social information to provide the personalized search and improve the result quality. Fourth, for the continuous mobile search queries, we propose effective indexing structures and devise efficient search algorithms to avoid unnecessary repeatedly queries, and we aim to reducing network transmission and battery energy consumption of mobile phones.

随着移动网络的发展和智能手机的普及，移动搜索作为一项新兴技术，受到了越来越多的关注。基于位置的服务是移动搜索必备功能，据统计95%智能手机用户使用过基于位置的服务，50%地图搜索来源于移动用户，因此基于位置的服务得到了广泛应用。针对传统基于位置服务的方法存在的不足和缺陷，课题研究多维度空间位置信息获取与智能检索，提高检索质量，改善用户使用体验，具体研究内容包括：（1）多维度空间位置信息的获取与整合：针对传统地图数据只包含简单有限的空间信息，研究多维度信息的获取与整合，提高地图数据的质量；（2）空间和文本信息融合的智能检索：利用多维度信息，深入融合空间和文本信息，提高检索质量；（3）空间和社会网络融合的个性化检索：利用社会网络信息为用户提供个性化、精准化检索服务；（4）面向移动对象的连续关键字查询：针对移动用户的连续查询，研究高效的索引和算法来减少不必要查询，降低网络传输和手机电池能量消耗。

项目按照原计划执行，解决了提出的科学问题，完成了相关研究内容，取得了多项创新成果。首先解决了多维度空间位置信息的获取与整合问题，提出了基础众包和签名技术的方法能够高质量、高效率的实现空间位置信息的融合问题。其次，解决了空间和文本信息融合的智能检索问题，提出了空间文本融合的topk检索方法，为空间文本数据检索提供了新技术。第三，解决了社会网络和文本信息相融合的个性化检索问题，提出了社交和空间文本融合的三维检索方法，将社交维度引入到时空检索问题中，提高了检索质量。最后，解决了面向移动对象的连续关键字查询，提出移动物体位置预测以及查询方法，能够实现高效的连续查询。发表计算机学会A类论文20余篇，相关成果应用到神州专车系统中。培养了10余位高水平学术创新骨干。