How Social Media AI Improves Disaster Response
Social media communications are obviously of critical importance during emergency situations. This communication channel allows utilities to convey important information to the public during emergencies, and it also allows the public to alert utility companies of anything amiss in their neighborhoods. Of course, this constant two-way barrage of information can create a lot of noise, and that is why there is a new push for using social media AI to make sense of it all.
What Exactly is Social Media AI?
Social media AI (artificial intelligence) uses machine learning to, as the phrase implies, learn over time what constitutes relevant posts in different situations. Machine learning enables an algorithm to identify patterns and correctly categorize bits of information. This allows the algorithm to funnel the information and summarize only what is relevant.
The first step before this can happen is to “train” the algorithm so that it knows what to look for. To do this, human volunteers look at various tweets, posts, chats, etc., and tag each one to categorize and measure the relevancy of the information. The algorithm essentially shadows this process to ultimately learn what to look for.
The possibilities of this are being explored in a multidisciplinary project to develop “human-AI learning” involving multiple universities and community agencies. The project was recently awarded a $50,000 grant from the National Science Foundation to continue with the research.
The overall goal of the multidisciplinary social media AI project is to develop systems that can scan and filter through millions of posts in real time, and harness the information to maximize situational awareness across the entire spectrum of interested parties. Understanding what is happening in a specific location can help emergency responders and utility restoration crews respond to the situation more efficiently and effectively.
The bottom line is that social media AI appears to be an effective tool for eliminating noise and reducing information clutter, to identify the most relevant information and categorize it for use in the field. Hopefully usage of such a system will expand dramatically in the coming years.
