Each week we will bring you a summary of what happened this week on our site, on Twitter, and in the wider world of civic data. Suggest stories on Twitter with #ThisWeekInData.
CityLab discussed the use of facial recognition technology in three stations within São Paulo’s Metro system. Newly-installed sensors monitor commuters’ reactions to posted advertisements and classify them as happy, unsatisfied, surprised, or neutral. They also track the quantity, age, and gender of viewers. The new software has inspired concerns about privacy, security, and ethical justification for personal data collection.
The Center for Data Innovation reported on the newest version of Google’s Open Images dataset, a nine-million-image public dataset for use in improving computer vision models. Images span across a broad range of object classes and complexities and contain bounding boxes, largely manually drawn to ensure accuracy, and object location annotations.
RouteFifty examined the second session of the IBM Center’s “Envision Government in 2040,” which intends to help governments prepare for significant changes to their work over the coming decades. At the second session, which focused on the future of AI, speakers touted potential advantages of AI such as time savings, customization, real-time trend assessments, and backlog reductions, but also noted concerns such as biased underlying data, the ethics of making certain decisions through AI, and developing leadership to advance IT systems.
Benjamin Clark and Jeffrey Brudney published a study investigating the diversity of callers into San Francisco’s 311 service request system, finding that citizen representation did not vary systematically according to a large set of socio-economic factors. A lack of bias in 311 services enables equal service to all residents, and quicker responsiveness may stimulate more active citizen participation, promoting a positive feedback loop for government services and citizen engagement.
Here on Data-Smart, Laura Adler analyzed the ways in which local governments and organizations are incorporating data to achieve better outcomes for disadvantaged children. The article highlights how detailed analytics help workers predict good locations for foster parent recruitment, assist foster families, and proactively identify and support children at risk. Mining data has also proven effective in improving programs for kids who have encountered the criminal justice system, which can significantly reduce recidivism rates. In all cases, Adler recommends that governments track project outcomes to determine paths to success and pool resources with engaged private organizations.
The Center for Data Innovation also announced Dartmouth College’s publication of Hypertools, a package that transforms datasets into visually accessible 3-dimensional static images or animations. The program provides a variety of tools for manipulating and visualizing large, complex data stores, which boosts users’ ability to identify trends and may assist in the training of machine learning algorithms.
The Upstate Data Project announced their upcoming Upstate Data Summit, to be held at Syracuse University on June 6th. Speakers at the event will showcase innovative technology and data-driven projects occurring in upstate New York and offer advice to officials looking to pursue similar programs in their own municipalities.
Radhika Garg published an article through the University of Illinois at Chicago about open data privacy and security policy. Garg attributes to stakeholders the responsibility for incorporating IoT into daily life, but examines the risks and barriers to the success of increased connectivity. According to Garg, IoT will only become ubiquitous if it avoids making users "prone to non-negotiable risks” such as a “lack of standardized security and privacy measures.” Thus, organizations and users alike stand to benefit from standardized systems for protecting information.
Wired reported on the Trump administration’s meeting this week regarding plans to aid in the development of AI. The meeting brought representatives from a wide variety of industries together to discuss implications of AI, particularly in terms of economic growth and job displacement. Michael Kratsios, Deputy U.S. CTO, revealed White House plans to begin sharing some government data with companies and to establish a new committee to help governments make the most of AI.
Co.Design examined the eye-opening relationship between popular music and economic conditions. Researchers at Claremont Graduate University analyzed lyrical content from top-100 music lists and classified it based on sentiment. They found that lyrical data tends to follow economic trends; for instance, songs released after the market crisis of 2008 expressed more disgust and anger and less trust. This relationship may carry policy implications, as the chief economist at the Bank of England has encouraged employees to examine current music tastes as a litmus test in determining whether or not to raise interest rates.
Co.Design also profiled a new interactive data illustration by Justin Fung. The visualization presents the population density on every New York City block by time of day for every day of the week. Both height and color of a given block serve to illustrate the number of people on that block at a given time. The financial district, for instance, becomes quite crowded during work hours and much less so at night. Fung believes that tracking population migration throughout the city may help the city plan for emergencies.