Using the GDELT Dataset to Analyse the Italian Sovereign Bond Market.

摘要: The Global Data on Events, Location, and Tone (GDELT) is a real time large scale database of global human society for open research which monitors worlds broadcast, print, web news, creating free platform computing the entire world’s media. In this work, we first describe data crawler, collects metadata GDELT in real-time stores them big management system based Elasticsearch, popular efficient search engine relying Lucene library. Then, by exploiting engineering detailed information each news encoded GDELT, build indicators capturing investor’s emotions are useful to analyse sovereign bond market Italy. By using regression analysis power Gradient Boosting models from machine learning, find that features extracted improve forecast country government yield spread, relative baseline where only conventional regressors included. improvement fitting particularly relevant during period crisis May-December 2018.