Mine Social Media and Other Online Data as Part of an Overall Ongoing Initiative to Better Understand Community Concerns.
- Watch residents’ questions for trends (5A)
- Track hashtags and trending topics (5B)
- Use social media data and location to understand where concerns are arising (5C)
- Use social media analytics on one’s own sites (5D)
- Consider future use of predictive algorithms (5E)
- Create data mining policies/ethical standards and training (5F)
WHY IMPLEMENT THESE IDEAS?
As a way to keep track of concerns raised by groups within their communities, and the depth of their feelings about them, community leaders can augment their advisory councils, community relations commissions, and constituency reports with policies and practices to mine social media data (sometimes called data analytics). Developing these policies and practices with the involvement of a stakeholder group can avoid both the reality and the perception that data mining will interfere with users’ privacy or be used to chill speech or assembly.
Social media mining varies. It may mean merely watching the trending information already available on social media platforms. At the other end of the social media mining continuum, it might involve the use of powerful algorithms that combine social media data with geography, research on human behaviors, and other factors. The mining may also range from noting existing concerns to predicting future actions. Social media is a powerful data source for research.
COMMON WAYS TO LEARN ABOUT TRENDING CONCERNS INCLUDE:
On most major social media sites, users can discover topics that are popular within their communities through the websites’ use of trending topics lists. Both Facebook and Twitter have created algorithms that keep the trending lists up to date and specific to a user’s interests and community in which they live.
Hashtags symbolized by the “#” symbol are used to group posts about the same subject together in one place. Originally created for Twitter, hashtags have since spread to most major social media sites. Hashtags have become an important tool for activists and others concerned with issues within their communities. The hashtag #BlackLivesMatters was one of the most tweeted hashtags of 2015 and was central in starting and maintaining the overall movement.
Most social media sites allow users to include their location when creating a post. In addition, social media allows for quick surveys about concerns. For example, the New York Police Department recently began sending 50,000 surveys per day on several smart phone applications. The surveys ask about a sense of safety and trust for police. A private vendor reports anonymous precinct results to precinct commanders, along with historical data so that they can compare results after new initiatives to serve their communities.[4a]
Mapping users creates controversy though if users suspect that public officials track them in order to dampen their ardor to protest. In 2016 during the North Dakota Pipeline Protests, users believed that public officials were tracking those who were communicating from the protest site to identify them. In response to this story, users around the world began reporting falsely that they were at the protests to thwart what they thought were the authorities’ use of geographic location to suppress speech and assembly rights.[4b] Although this rumor was false according to a statement released by law enforcement on Facebook, it increased the distrust between the protesters and law enforcement and made legitimate police work related to the protests more difficult. Nextdoor allows public use of location without also allowing disclosure of individual identities, permitting monitoring of concerns without leading to residents’ concerns about repercussions for raising them.
It may be simple to use social media analytics on one’s own sites (followers, posts, and related information) and such an approach will help to connect with community members and tailor messages to build community relations. On Facebook, for example, users with professional pages can use the Insights tool to learn the age, gender, location, and language spoken by people who “like”. They can find out how many people have viewed a post or followed a link, and other useful pieces of information. Users can learn the reach and performance of specific tweets on Twitter and some basic information about the people who follow the account and their interests. Instagram offers similar analytic tools for professional users (not individual accounts). Community leaders can learn about topics of interest, when to post important messages, what formats can reach the largest audience, and more.
Some services use social media data to analyze community problems and to predict the next occurrence of civil unrest. Through these predictive analytics programs, communities can identify issues that are gaining traction so that they reach out during tranquil times to convene constructive discussions about the issues that might lead to change.
One predictive model that has found success in Latin American and has been used recently in the Middle East and North Africa is the Early Model Based Event Recognition using Surrogates program (EMBERS). Though not available in the U.S. as of this writing, research on EMBERS shows the capacities of such complex predictive models that use publicly available data including social media data and economic data among other sources. EMBERS predicted violent incidents in Brazil and Venezuela between six and 11 days before they occurred.
5F. Local leaders will want to accompany any data mining with transparent policies/ethical standards that take into account the public’s likely fears concerning privacy, free speech and assembly, and discriminatory use of data.
In other words, using data to learn more about the community to build trust and to improve community relations risks doing just the opposite if local leaders do not develop transparent policies that take these concerns into account. Though the data may be publicly available, the public often responds negatively to new ways of using it though data analytics, especially when done by the government.
Concerns seem to mount if the data mining involves:
- Use of a predictive model that will negatively affect a particular individual. For example, people may oppose use of social media data to predict behavior, such that law enforcement questions that individual. People have criticized the legal use of public data through algorithms such as programs that determine the price of bail for an accused individual.
- Use of algorithms that affect people differently (and negatively) on the basis of race, gender, or income levels. Although the algorithms have proven that they are important factors in making determinations, such as predicting civil unrest, the use of factors such as race, gender, and income level produce risks of discriminatory outcomes if applied to whether a particular individual is protesting or likely to protest. If these mathematical prediction models do not account for societal issues such as racial inequality or problems associated with poverty, they can result in outcomes that are viewed as inherently unjust.
- Use of data to stifle protests. The American Civil Liberties Union (ACLU) has warned that the practice of tracking protesting individuals in an effort to stifle protests is an unconstitutional action that violates individuals’ First Amendment rights to free speech and assembly.
 Globally—as measured at 11:10 am on March 31, 2017—each second, 7,562 Tweets are sent, 774 photographs are posted to Instagram, and 68,702 YouTube videos are viewed. Further, that data can be combined with material from other sources. Each second, for example, users conduct 59537 Google searches. Internet Live Stats, http://www.internetlivestats.com/one-second/#tweets-band (last visited Mar. 31, 2017).
 Seattle Channel, http://seattlechannel.org (last visited Mar. 31, 2017).
 Tanya Sichynsky, These 10 Twitter Hashtags Changed the Way We Talk About Social Issues, The Washington Post, The Switch (Mar. 21, 2016) (ranking the “10 most influential hashtags around social causes, ranked by the number of times they’ve been used since their inception” as provided by Titter).
[4a] Al Baker, Updated N.Y.P.D. Anti-Crime System to Ask: ‘How We Doing?’, New York Times (May 8, 2017).
[4b] Merrit Kennedy, More Than 1 Million ‘Check In’ On Facebook to Support the Standing Rock Sioux, NPR (Nov. 1, 2016).
 Morton County Sheriff’s Department, Facebook (Oct. 31, 2016),.
 Twitter user or Instagram user who has subscribed to an account in order to view the account’s Tweets or pictures in their timeline.
 To like something on social media is originally a Facebook feature that evolved into an understood expression of support for content across platforms. You can “like” an Instagram post, for example. Along with shares, comments, and favorites, likes are tracked as proof of engagement.
 Andy Doyle et al., Forecasting Significant Societal Events Using the Embers Predictive Analytics System, 2(4) Big Data 185, 186 (2014).
 Kerry Flynn, Social media companies suspend Geofeedia’s access after reported policy tracking, Mashable (Oct. 11, 2016); Nicole Ozer, Twitter Cuts Off Fusion Spy Centers’ Access to Social Media Surveillance Tool, ACLU (Dec. 15, 2016); Kate Groetzinger, Slacktivism is Having a Powerful Real-World Impact, New Research Shows, Quartz (Dec. 10, 2015). See also Laura Seay, Does Slacktivism Work?, The Washington Post (Mar. 12, 2014).
 See Pablo Barberá et al., The Critical Periphery in the Growth of Social Protests, PLOS One (Nov. 30, 2015); Ting Hua et al., Analyzing Civil Unrest through Social Media, IAEE Computer Society, Nov 22, 2013, at 80; Andy Doyle et al., Forecasting Significant Societal Events Using the Embers Predictive Analytics System, 2(4) Big Data 185, 186 (2014).
 See generally articles cited in Note 9.
 See generally articles cited in Note 9.
 See generally articles cited in Note 9.
 Matt Cagle, Facebook, Instagram, and Twitter Provided Data Access for a Surveillance Project Marketed to Target Activists of Color, ACLU (Oct. 11, 2016); George Joseph, How Police are Watching you on Social Media, CityLab (Dec. 14, 2016).