Generating Automated News to Explain the Meaning of Sensor Data

Abstract

Abstract. An important competence of human data analysts is to interpret and explain the meaning of the results of data analysis to end-users. However, existing automatic solutions for intelligent data analysis provide limited help to interpret and communicate information to non-expert users. In this paper we present a general approach to generating explanatory descriptions about the meaning of quantitative sensor data. We propose a type of web application: a virtual newspaper with automatically generated news stories that describe the meaning of sensor data. This solution integrates a variety of techniques from intelligent data analysis into a web-based multimedia presentation system. We validated our approach in a real world problem and demónstrate its generality using data sets from several domains. Our experience shows that this solution can facilítate the use of sensor data by general users and, therefore, can increase the utility of sensor network infrastructures.

Proceedings of the 9th International Natural Language Generation conference

“…KDD community (Guillet and Hamilton, 2007) and employed lateraly in other NLG systems (Molina et al, 2011).…”

“…Similar to (Molina et al, 2011), we seek to summarize relevant facts with explanatory descriptions and graphical information. However, we have a different main goal which is to provide an overview of any tabular data without extra domain knowledge provided by the user.…”

Simulating Data Journalism to Communicate Hydrological Information from Sensor Networks

“…This is useful for certain sensor networks similar to the SAIH network. However, other networks (e.g., domains with moving sensors) can require different representations (e.g., event-based approaches) [14].…”

Latent User Models for Online River Information Tailoring

“…Explicit models then define a finite number of user groups, and finally generate tailored texts for users to choose from, or choose to generate for a unique group at each time, e.g. (Molina, 2011 and2012). Implicit models, e.g.…”