A Knowledge-based Restricted Problem Solving Method in Gis Applications

What this paper tends to address are followings. (1) A notable common characteristic of many GIS applications we ever might ignore is that the ability of analysis is not primarily provided by GIS itself, GIS generally plays roles of data management and virtual environment construction, and it is the users of GIS software that achieve that intelligencerequired process of problem solving. That is to say the power of analysis is comparative weak in classical GIS. The main function of classical GIS was confined in data integration or data fusion through objects’ positional correlations. (2) Why not let GIS execute more tasks needing complicated analyzing like an active advisor in stead of a passive demonstrator? Or how to make traditional GIS more power in complicated problem solving? After all this is a significant and challenging problem both in GIS and Artificial Intelligence. (3) The aim of this paper is to develop a systematic method to combine GIS and AI together, thus make an enhanced GIS and enable it exhibits intelligence as more frequently and more automatically as possible. The key difficulty this idea concerning is the usage of multi-domain knowledge and meta knowledge. So modules of reasoning, planning and decision-making are very necessary in an actual GIS-based application. (4) Many GISbased applications seem to be simple at first glance, but if what you want to seek is not only a demonstration for principle but a comparative practicable system, then a difficult problem is put forwards. Because from the point of view of Artificial Intelligence this problem solving process is domain-cross, domain-open and common sense-required, all of them are the research fronts of Artificial Intelligence. In general a deep intervention of Artificial Intelligence into GIS will greatly promote the development both in the applications of GIS and also in the theories of Artificial Intelligence.