摘要
针对基于位置的web搜索需要将网页中位置信息和文本信息相结合进行索引的应用需求,提出了先倒排表再R-tree索引和先R-tree再倒排表索引两种混合索引结构,同时处理文本和位置信息.大规模真实数据集上的实验表明,这两种方法在查询效率上明显优于已有的倒排表和R-tree相互独立的索引模式.
For location-based web search, geographic information should be indexed with textual information. Two hybrid index structures were proposed to deal with both textual and geographic information： one of inverted file preceding the R-tree and one of R-tree preceding the inverted file. Experiments on large real-world web datasets show that the proposed structures have better query performance than the existing index schema of separate inverted file and R-tree.
基金
国家自然科学基金青年基金(604030200)资助
