A Study of Automatic Recognition Mechanisms of Meanings of Location Names in Document Databases

在之前的工作中，我们开发了一个上下文相关的地理编码系统。地理编码系统设计用于将位置名称转换为相应的地理编码。我们的地理编码系统的主要特征是通过处理位置表示的空间上下文来识别位置表示的含义。我们将空间语境定义为与歧义位置表征的单一含义强烈相关的一组术语。我们的地理编码系统是使用一个特殊的元数据库来维护地标的空间上下文。在这项研究中，我们开发了一种新的元数据库生成方法。该方法的主要特点是通过处理地标名称和特征词在文本数据库中的共现来提取地标的空间上下文。因此，我们的地理编码系统可以应用于各种文本数据库。通过多次实验，验证了该方法的可行性和有效性。

We developed a context-dependent geocoding system in our previous work. Geocoding systems are designed for translating location names to corresponding geocodes. The main feature of our geocoding system is to recognize the meanings of ambiguous location representations by handling the spatial contexts of the representations. We define a spatial context as a set of terms strongly relating to a single meaning of an ambiguous location representation. Our geocoding system was implemented to use a special metadatabase maintaining spatial contexts of landmarks.In this study, we have developed a new implementation method for generating the metadatabase. The main feature of our method is to extract the spatial contexts of landmarks by handle the co-occurrence of the landmark names and feature words in a text database. Thus, our geocoding system can be applied to various text databases. We have demonstrated the feasibility and effectiveness of our method through several experiments.