Generating text descriptions for geographically distributed sensors

Abstract

Sensor networks, with thousands of geographically distributed sensors and different types of quantitative measures, need software tools to help users understand the meaning of measures. In this paper we pay attention to the problem of automatic generation of geographic descriptions in natural language for geographically distributed sensors. We describe this problem in the context of a web application in the domain of hydrology which is part of a more complex multimedia presentation system that combines text and graphics. We describe the web application and the algorithm that we designed to generate the geographic descriptions for sensors. Besides GIS data files, our method uses two information sources: an online server for geographic names (Geonames) and a specific knowledge base with text patterns that we constructed to process sensor identifiers. The evaluation results confirm that online geographic information resources such as Geonames are useful to generate names for sensors but they need to be combined with other more specific information sources (such as our knowledge base) to obtain good descriptions. We also compare our method with related work and show future lines of work.