Sentiment Analysis for COVID response​

Autor: Yakov Ozer (Microsoft)

COVID-19 Crisis Response requires Governments and Healthcare Providers to adopt an aggressive social media channel engagement strategies for Citizens and Entities so they can:​

  • Promote the latest “authorized” guidance​
  • Provide consistent engagement across platforms​
  • Amplify through followers the consistent messaging​

Sentiment Analysis will be required:​

  1. Understand which messages/communications landing well​
  2. High level feedback look concerns/feedback to update guidance ​
  3. Key influencers to engage​
  4. Negative sentiment and or even “fake” advice​

The results are:

  • Publish a report of Public sentiments at EU, Country, Region level by reading Twitter feeds and hashtag text​
  • Create a Word cloud of most frequently used emotions in the feeds, using Term Frequency, Inverse Document Frequency​
  • Create a real time dashboard in Power BI with public emotions on Twitter, arising from COVID and associated hashtags​
  • Calculate the overall sentiment score using keywords, Positive, Negative, Neutral​
  • Sentiment breakdown into various different emotion types​
  • Find out any live insights, escalation of emotions across various Countries, Regions​
  • Progression of Sentiments over a period of time, to assess the severity of the situation on ground.​
  • Care service level sentiments of various Health agencies, any support cries or any immediate actions required from agencies​