Tom Sabo is principal solutions architect at SAS.
This year’s presidential election was like no other, and American citizens continue to voice their opinion on all things government, most notably on how it can improve.
A recent Pew Research Center poll revealed 80 percent of citizens do not believe the federal government runs its programs well, while another 59 percent believe government needs major reform. With the next president, one might assume a new leader could spur that change, but 74 percent of Pew respondents believe elected officials put their own interests ahead of the country’s.
Whether you agree or disagree with these findings, they provide a pulse on American sentiment at this moment time. Government continually requires this kind of honest insight. However, polls like this are time-consuming, expensive and offer only a narrow view of public opinion, which may or may not be actionable. To affect real change, agency leaders need a deeper and more consistent look into citizen views.
Turning to Advanced Analytics for Text Analysis
In the age of social media, text and sentiment analysis has become an indispensable capability. Through publicly facing social media sites as well as forums and blogs, citizens leave a public record of their likes and dislikes, whether it be a picture of a delicious dinner at a restaurant or an angry tweet sent to an airline. Text and sentiment analysis can derive actionable information from this text-based data. Advanced analytics tools can rapidly identify and visualize more accurate information, and deliver it faster than traditional survey methods.
Commercial organizations use this kind of feedback every day to improve the goods and services they offer to their customers, and ultimately, their bottom line. Government organizations can as well, in a genericized manner and where permissible by law, use the same data to improve services offered. This also includes consumer protection as well as emergency management use cases.
Inside the Technology
Advanced analytics can mine unstructured data to gauge public opinion and discover overarching trends. This is done while maintaining privacy—the analysis does not identify the person who said it, it simply focuses on what was said. With the right tools, government agencies can track constituent sentiment to drastically improve operations. The end result could resemble that of a customer-focused business.
Roughly 90 percent of all data is unstructured and requires much more than manual analysis to put it to good use. Text and sentiment analysis can evaluate text in a natural fashion, similar to the human mind, by following sophisticated linguistic rules and analytic modeling.
With machine learning, users can model for and subsequently edit rules based on definitions and categories, ensuring the analytics tools interpret text using both information inherent in the data alongside of the subject matter expertise available to the organization.
Data can also be analyzed quickly, assuring agencies have timely and relevant data on which to base their decisions. This agility can fundamentally alter how agencies prepare for and respond to varying levels of constituent sentiment on particular relevant topics, such as those relevant to the U.S. economy or foreign policy.
The Consumer Financial Protection Bureau already leans on advanced analytics to contextually assess consumer complaints. CFPB collects complaints from consumers and then sends them to the companies themselves. Since its inception, CFPB has received more than 800,000 consumer complaints, almost all of which include a free-form textual description.
Along with sending those complaints to the company they were filed against, CFPB performs an analysis on their content. That data helps agency leaders understand the financial marketplace and better protect consumers. This is especially important when receiving complaints about businesses such as banks, money lenders and mortgage brokers that can have a large financial impact on consumers.
Advanced analytics and sentiment analysis hold a world of potential for other parts of government as well.
Imagine the government’s health agencies using social media posts to understand, track and fight infectious diseases. FEMA could use these tools to better target relief efforts during a crisis, similar to how the International Organization for Migration used text analysis of social media in the wake of Typhoon Haiyan in 2013.
Congressional leaders could potentially use text and sentiment analysis to analyze how their constituents will respond to specific legislation. Congressional leaders and their staff do not have time to read all the traditional, electronic and social media communication sent to them from their constituents. Even if they did, only qualitative interpretations result from solely manual analysis. Advanced analytics can help them to quantitatively gauge citizens’ thoughts on a topic. The end result is shorter time to value with more accurate understanding.
These types of applications are the tip of the iceberg. With advanced analytics, government can better meet its mission and address public needs, while saving on information gathering methods. The Pew Research Center poll and others like it provide great information about a moment in time. As time goes on, individual polls lose value. Analytics such as discussed here enable government to keep a regular pulse on its constituents, to best understand the changing needs and wants of the people.