I've heard a lot about text analytics since I started tracking the companies who monitor and analyze social media. I've been particularly interested in the different directions people go with the same basic source data and analytical techniques. It's not all sentiment and PR; once you start extracting opinions and themes from streams of online content, the next step is to look for correlations between online trends and external events. In marketing, you're probably looking for a link to sales; some of the companies I talk to are more interested in stock prices. But if you want to know what's really wrong with the world, you might start with Europe Media Monitor (EMM) from the European Commission's Joint Research Centre.
EMM pulls current world news from online sources, clustering reports by topic and tagging them with location and language. The real charm of the service is its optimistic focus on themes such as terrorism, communicable disease, political unrest and conflict. Clustering makes it easy to track current news on your specific combination of interests in the language(s) you can read. Of course, everything is multilingual, with "about 19" languages represented (the language filter is always there to keep things readable).
The result is a customizable news service available to government and private users through web-based dashboards, RSS and email. The site has three major components:
- NewsBrief, a quick view of the most active stories and themes, updated every 10 minutes. In addition to the global summary, NewsBrief has current news for 13 EMM themes, EC Directorates, selected international organizations, and a handful of topics focused on the European Union.
- NewsExplorer, a global news dashboard with drill-down analysis on nations, individuals and other entities. Detail ages show a breakdown of recent news by theme, plus excerpts of current articles.
- MedSys, a real-time dashboard and alerting system for medical and health-related issues. Detail views focus on specific diseases or bioterrorism threats, along with a volume trend chart and interactive map for drill-down analysis. In addition to email, MedSys can send alerts by SMS.
Registered organizations (governments?) get additional filtering and reporting options. There's also a restricted version of the MedSys site with additional information.
If you're a news junkie, I'm sorry. I didn't mean to throw your productivity out the window, but this is worth checking out. The question I'm finding most interesting lately is, "what else is this good for?" Given the history of interest in text analytics by intelligence and law enforcement agencies, it's no surprise to know that they're still looking for ways to predict the future. It is a surprise to find such interesting sources available to the public.
Hat tip to Penny Herscher, who pointed out the article from Wired.
