The Geoimage Generation Method for Decision Support Systems Based on Natural Language Text Analysis

Abstract

AbstractGeospatial data is a critical component of decision support systems for regional governance. The use of spatio-temporal data is most useful for rapid mapping of socio-economic phenomena and analyzing changes in the main indicators of regional development. An important advantage of using geospatial data is its versatility and applicability to a wide range of decision support tasks. Many modern decision-support information systems have their own tools for geospatial data processing. Where such tools are not available, integration with geographic information systems for visualization and analysis of geospatial data is possible. An important factor is the speed of obtaining and visualizing geospatial data. This paper describes the rapid geoimage generation method for regional governance decision support systems based on the analysis of natural language texts. The first part of this paper describes the structure of a system for recognizing named entities from texts in Russian, based on hand-crafted grammars and gazetteers. It also shows a fragment of the grammar and describes the method for populating the gazetteer. The second part of this paper briefly describes the process of geocoding and visualization of named entities extracted from natural language texts. The result of the method is an interactive geoimage (digital map). This geoimage is used by a decision-maker as an additional source of information for decision-making. A special feature of the proposed geoimage generation method for decision support systems is the high speed of obtaining geoimages displaying the named entities recognized in the text.KeywordsGeovisualizationDecision support systemsNatural language processingNamed entity recognitionGrammarGazetteerGeospatial dataRule-based approach