Metadata Design towards the Data Selection in the Integration of Geographic Information

• Main Author: 李寶玲
• Other Authors: Jung-Hong Hong
• Format: Others
• Language: zh-TW
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/12584455923276717969

碩士 === 國立成功大學 === 測量工程學系 === 87 === In recent years, the development of National Geographic Information System (NGIS) has established a lot of databases that provides naive users a chance to access various geographic information in a user-friendly environment. However, We argued that the integration of geographic information from different domains is much more than transforming them to the same coordinate system and data format, and then ‘integrate’ through map overlay process. The reason is simple: the data layers may be created in two totally different ways and may not be compatible from many perspectives. By carelessly considering them as the same level, the result of data integration most likely will be very risky. Therefore, we have to overcome the following problems before integrating geographic information from various domains: 1. Understanding the content of integrated geographic information. 2. Selecting data that is compatible via professional knowledge. Various related researches in the past suggested a metadata-based approach to aid the browsing, management and search of geographic information. The goal of this research is to deliberate about metadata design that is needed during data selection procedure in an integrated geographic information application. By introducing knowledge rules that provide constraints on integrated data, users are able to build correct understanding towards selected data. The mechanism of metadata design is based on the analysis of the characteristics of spatial data, especially those about the coverage, temporal and positional accuracy. The result shows that we can therefore distinguish among these situations: ‘the phenomenon exist and the data were recorded’, ‘the phenomenon exist but no data recorded’, ‘the phenomenon doesn’t exist but the data were recorded’, ‘the phenomenon doesn’t exist and no data recorded’. In spite of the fact that this paper may not detect all possible errors, yet it can improve the understanding of the integrated data and reduce the risk of lack of professional knowledge. We expect that metadata approach to expand from an ‘informative’ role to a more aggressive ‘data use’ role. For a distributed data environment like NGIS, such research will serve as a necessary component in the future.