This is a supplement code for the research paper dedicated to extraction of valuable terminology / facts from disaster-related news in Ukraine. The paper: Usage of Disaster News Information for Predicting the Emergency Situations in Ukraine
Script provided in Disaster prediction.ipynb:
- collects news from one of the Ukrainian internet news portals (24tv.ua)
- classifies them using provided tags
- extracts location where disaster happened using NER model for Ukrainian language
- detects anomalies in the news frequency by location
- compares number of gathered news with information published by State Emergency Service of Ukraine
News were collected from the web-portal of Ukrainian news - 24tv.ua
Database of the collected news is stored in the allnews.csv
Links to the news are stores in the urls_.csv
- scrapy
- NER model for Ukrainian language (https://github.com/lang-uk)
- numpy
- pandas
- scipy
- sklearn