A Preliminary Study Of GIS Semantic Mining Based On Requirement Text

Semantics is traditionally defined as the study of meaning, which includes linguistic meaning and symbolic meaning in logistical formalism. Form the perspective of semantics, GIS is a symbolic system in the logistical formalism, and GIS semantics can be understood as spatial conception represented by the GIS. The study of semantics plays significant role in different fields, and the traditional semantic theories and methods have influenced the study of the computer semantic and GIS semantic. In terms of semantic history, semantic mining has undergone several phases: small-scale manual mining from language; large-scale automatic mining from semi-structured and structured data; automatic mining from the special database (such as spatial database) .People have gradually realized the necessity of combining the unstructured text with the semantic mining methods in order to help people learn more about the implication of text from different aspects and to find the best mining location and gain more cognition.During investigation of GIS project, we collect much unstructured user requirement text .The text is systematized one on small scale and contains much GIS semantics, which is often ignored by people. Using semantic mining theories, methods and the experience in the semi-structured text and spatial database mining, we can get the GIS semantics from the user requirement text. It will help us improve the quality of requirement engineering and it is important in exploring GIS semantic sharing, GIS explanation standard and developer-application sharing .Additionally it gives people some new ideas about GIS mining from unstructured text.Firstly the thesis focuses on introducing the process of the formation of requirement text and analyzing its contents, semantic character and the main task of text mining in order to help people collect more valuable user requirement. Secondly the thesis deals with the spatial cognition of geographic environment, the concept of semantic and GIS semantic field, and summarizes the contents of GIS mining, the mining framework and the mining principle. GIS semantic mining can be defined in both broad and narrow sense .In a narrow sense, GIS semantic mining focuses on the spatial conception mainly mining form the oriented-object model, field model, GIS operation and GIS scenes. In a broad sense, GIS semantic mining concentrates on the GIS project, and refers to mining from GIS, software requirement and business rules. It mainly mines the GIS information communities, shared manner, glossary, metadata service, the context of GIS, the boundary of GIS product, the field model of GIS, the business model, the functional requirements and non-functional requirements. Finally based on some user requirement text in National 863 project (Environmental pollutant Accident and prediction System), the thesis presents the special examples of GIS semantic mining, and proves the feasibility and significance of GIS mining based on requirement text. It also paves the way for the further study of automatic mining of GIS semantic from infinite amount of text.