Abstract: Knowledge service is an important application direction of GIS. The analysis and mining of the rich implicit geographic information contained in massive text data has become a hot research issue. In the field of natural resource management, the distribution of natural resources within a certain temporal and spatial range is relatively independent and scattered. The rich semantic information in the text is fragmented, complex and highly unstructured, lacking effective organization, integration, and comprehensive application solutions. Oriented to text data and natural resource geographic entities, this paper proposes a semantic-driven geographic entity expression framework. It organizes and expresses the multi-domain information of geographic entities through a four-tuple of semantic description, spatial location, attribute characteristics, and temporal evolution. It defines and describes the multiple types of relationships between entities from the four dimensions of concept, space, attributes and time. Following the steps of geographic entity information extraction, information storage and association construction, we give a method for constructing a multi-dimensional geographic entity association network. Then, we design a knowledge question answering algorithm based on the associated network. Finally, taking construction land approval as an example, using electronic text data of the approval process, we complete the materialized expression of construction land information, the construction of the geographic entity association network, and the realization of knowledge question answering service. The experiments and analysis show the theories and methods of this article can effectively promote the organic integration, full association and scientific management of geographic information in the text, and provide practical ways to improve application and social service level of information in the field of natural resources.

Key words: text data, semantic-driven, geographic entity expression framework, association network, knowledge question answering service