The data
In NCSR “Demokritos” – a partner of the AI4EU platform – we employ AI to understand the communication on the web, concerning specific entities and settings (Social Web Observatory service). We also possess the tools and methods to summarize information from unstructured text, by using natural language processing (NLP) technologies and text mining techniques (NewSum toolkit). The analyzed data form an excellent source of pre-compiled, soft data, which can be used in many ways, such as:
• To understand the public sentiment for a stated policy, an action plan, a campaign.
• To recognize or prognosticate threats and opportunities.
• To help to deal with misinformation and disinformation.
The services
The services provided include:
• Mining communication channels in the public sphere to identify epidemics-related statements
• Identifying context-indicative meta-data in this communication (e.g. location names, legal entities involved, virus/disease names) through NLP, to enrich the free text with added value
• Understanding the expressed sentiment in the analyzed messages, concerning specific entities if appropriate
• Identifying groups/clusters of communication items (messages/posts/news) referring to the same real-world event
• Summarizing and visualizing the events and their related information in an actionable manner
• Allowing real-time analytics, which can help in-depth analysis of analyzed documents
• Create a chatbot for Q&A for COVID-19. This could provide useful info or truthful facts regarding COVID. Also, regarding the government perspective, it would be able to answer questions regarding non-health-related stuff or at least link to the official relevant sources (π.χ. Updated travel ban for Easter, site for the forms (forma.gov, etc), 600/800E voucher for freelancers, etc.)
• Due to recent cases where people infected are asked to stay at home but do not get clear instructions between their personal doctor and EODY, create a structured QA form like NHS has done. For example, more than just info on what to do would help if there is an urgent need for a patient to be transferred to hospitals. We also gather geospatial info with this.
Examples
Epidemics-related
• It is needed to classify sets of false claims for “drugs that can deal with the virus” or related conspiracy theories and deal with them.
• It is needed to identify how the citizens perceive the free movement restrictions and whether there exist specific problems in the related bureaucracy, to facilitate them.
• The aim was to understand what impact the “stay at home” campaign had, to follow up on more specific audiences.
• Another goal was to investigate the effect of movement restrictions on people’s life/health/mentality.
B2G related
• We can improve citizen engagement by gaining intuition from public opinion and response, for policies, actions, campaigns, etc.