顾及地理实体的地名信息检索方法研究

随着在线电子地图应用的普及,地名信息检索已成为相关研究领域的热点,并取得了丰富的研究成果。然而,这些研究主要侧重于检索效率和文本关键词匹配质量的提高,并未顾及实际应用需求和地名信息检索方法的矛盾。本文通过分析地理实体和地名的特征及关系,提出将二者相融合来研究地名信息检索的形式和方法,从数据特征和应用场景的角度出发,提出顾及地理实体的地名信息检索模式,并实现对其的结构化表达,在该检索模式中将地理实体的属性特征和空间关系作为检索特征项;以检索模式为基础提出了基于地理实体相似度计算的地名信息检索模型,并深入研究了文本型属性与数值型属性综合的属性相似度计算方法,以及定性空间过滤与定量距离计算综合的空间关系相似度计算方法;最后,通过实验验证了顾及地理实体的地名信息检索方法的可行性和检索质量。实验结果表明,该检索方法既能实现精细化和空间化的检索需求,也能保证检索结果的准确度。

A Place Name Information Retrieval Method Considering Geographic Entity

With the popularization of online map application, place name information retrieval has become a key issue in the related study areas and many promising findings have been achieved. However, these studies mainly focused on the efficiency of retrieval and the matching quality of text information, without considering possible contradictories between the practical application requirements and place name retrieval methods. Currently, many online map platforms generally take POI or point of place name as the retrievable objects, and get results through retrieving text key words. However, this approach ignores the spatial information of place name. This paper studies the forms and methods of place name information retrieval by combining geographic entity and place name based on analyzing the relationship between them. From the view of data feature and application scenes, a new place name information retrieval mode is proposed, which can fully express the structure of geographic entity. This retrieval mode takes geographic entities' attributes and their spatial relationship as the retrieval feature items. With this mode, this paper proposes a place name information retrieval model based on geographic entity similarity. In the meantime, the similarity calculation method for attribute that integrates text attribute and number attribute, and the similarity calculation method for spatial relationship that integrates qualitative spatial filtration and quantitative distance calculation are also presented. Finally, a series of validation experiments are conducted to test the practicability and accuracy of the proposed model. Results show that the retrieval method not only considers the details and spatial characteristics of the retrieval objects, but also ensures the accuracy as well as the quality of the search results.