Macroeconomic forecasting with statistically validated knowledge graphs

作者：

Highlights：

• Features derived from graphs based on news themes improve predictions of economic activity.

• Disease and economy-related themes have the strongest predictive power.

• Such graphs can be modified to track specific events in real-time and thus have many applications.

摘要

•Features derived from graphs based on news themes improve predictions of economic activity.•Disease and economy-related themes have the strongest predictive power.•Such graphs can be modified to track specific events in real-time and thus have many applications.

论文关键词：Knowledge graph,Time series forecasting,Natural language processing,Big data

论文评审过程：Received 21 April 2021, Revised 14 July 2021, Accepted 11 August 2021, Available online 18 August 2021, Version of Record 4 September 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115765