Text analytics is a new horizon for data-driven decision making in municipal government.
Why text analytics? If there's one thing local governments have in plentiful supply, it's text. Records of everyday government activities are loaded with it -- police reports, court proceedings, license applications, school records, and other documents from routine government business generate a sea of text.
Text sources can be rich in valuable information that does not exist in any other form. Consider the gems found in the notes of police reports -- details about weapons, behavior, past encounters with suspects or victims, and so forth. Locating relevant information in these records is a challenging process that has depended on memory and manual review in the past.
Now police officers in Alexandria, Va. -- a Washington, DC, suburb with a population of 150,000 -- are using text mining to track gang and drug-related violence. Text mining facilitates the identification of key concepts within things like police and crime reports, tips from the public, and news stories to help officers conduct research quicker and easier. Faster and more thorough searches mean better information for detectives and patrol officers.
Chicago city limits
Large cities are not about to let themselves be outdone by the suburbs. Chicago, for example, began exploring Twitter analysis to complement its 311 system more than a year ago. It now mines 311 calls to predict and prevent rat infestations.
Municipal 311 systems are fertile ground for text analytics opportunities. Records of past inquiries provide a tremendous resource for learning the kinds of information constituents want, what difficulties they face in finding it, and what their most frequent questions are. Manually organizing that information from thousands or millions of records is tough, but text analytics makes the process faster and more consistent.
Back to the roots
The exploration of text analytics by city governments is not just a US phenomenon. Hong Kong is using text analytics for its call center and paying particular attention to complaints, which represent about 10% of those calls. Why complaints? They carry hidden messages about root causes of problems with government services.
Identifying root causes is a unique value proposition for text analytics in government. It's one thing to know something happened -- a crime, a missed garbage collection, a school expulsion -- and another to understand where the problem started. Conventional data often lacks clues about causes, but text reveals a lot.
Chris Bowman, president of Educational Analytics and Logistics in Louisiana, explained to us how root-cause analysis works, using student expulsions as an example. "You would read that a student was in trouble for disrespect for authority, but the incident started as a question about not having a shirt tucked in, or a silly comment in class. How could a molehill become the Himalayas?" He realized that, if he saw this pattern several times in individual reviews, there would have to be more, but it was impossible for him to read tens of thousands of student records to find out. That's how he became one of the education industry's earliest adopters of text analytics.
These stories are just the beginning. Every city has problems, and every city has lots and lots of text. In the next few years, we'll be hearing many more tales of text and the city.
Is your city exploring text analytics? Do you wish it would? Share your stories.