Given the extraordinary volume of data emerging from social media, urbanists have a unique opportunity to gain enhanced insights into the attitudes and opinions of people in cities. While technologies are ever changing and ādisruption is the new norm,ā platforms like YouTube, Facebook, Instagram, Snapchat, and blogging demonstrate the breadth and durability of digital self-expression. Twitter, in particular, is among the most established and mainstream of microblogging venues. It is also inherently populist, with everyone from heads of church and state to heads of households, and individuals who make up those households, possessing Twitter accounts. It is the self-expression of this great breadth of individuals that we find most compelling, and our efforts here are geared toward making evident the ways in which research using large social media data sets provides an additional investigative tool for planners and policy makers, even as we provide a candid discussion of the ways in which this method might be improved.
Thus, this book seeks to show how Twitter, as a prominent case of digital self-expression, can be analyzed and contribute to our understanding of place. Moreover, this book explores the utility of a new and potentially robust data source: the Twitter Application Program Interface (API) feed. We provide some context for the broader use of social media in urban studies and planning, explore the strengths and weaknesses of data gleaned from Twitter, and then demonstrate how such data can be used in research and practice. The three demonstration studies were completed across more than a full year, with the work of Chap. 3 being conducted in the winter and spring of 2014, the work of Chap. 4 being conducted in the fall of 2014 and spring of 2015, and the work of Chap. 5 being conducted in the spring of 2015. Through these temporally and spatially distinct studies, the results provide varying answers and conclusions about this type of research. With this in mind, the book serves as a comprehensive methodological guide on how to collect, process, analyze, and interpret Twitter data and offers the first frank and honest assessment of the strengths and weaknesses of Twitter data.
Social media data sets, including data from the microblogging platform Twitter, constitute a part of what is known as āBig Data,ā a term first used by NASA in 1997 to describe quantities of data so large that they taxed memory and hard disk capacity (Cox and Ellsworth 1997). Since around 2008, the term has been used to describe a compelling phenomenon in academia, government, business, and the media in which data is no longer seen as an entity with limited value after an initial use; rather, it is an input to be used continually in innovation, service creation, and as a means of collecting information not previously available (Mayer-Schƶnberger and Cukier 2013).
In the past, massive amounts of electronic information, including personally volunteered data, were not nearly as readily available as they are today. Computing capabilities were comparatively limited, and data generated was static with specific and restricted significance. In contrast, Big Data is dynamic, continually relevant, and additive (and added to) and provides unique insights and opportunities not previously available (Mayer-Schƶnberger and Cukier 2013). This enables researchers like ourselves to test research questions that we would not have been able to meaningfully investigate even five years agoāat least not from the angles or with the input of millions of people that we can now.
There has been considerable recent attention in the academy regarding the potential applications and pitfalls of Big Data in efforts to better understand the social world (see Chronicle of Higher Education special issue in April 2013). This interest is, in part, a response to the difficulties associated with collecting attitudinal data using traditional survey-based approaches, which are prohibitively costly to conduct on a wide scale and which increasingly suffer from respondent fatigue and skepticism. Data collected through publically posted online social media platforms is far cheaper to collect, less obtrusive, and much more voluminous. For example, half a billion Twitter messages are sent every day. Twitter itself is the largest microblogging platform in the world, a type of instant message service that restricts users to public posts of fewer than 140 characters. Other microblogging platforms include Sina Weibo in China and ImaHima in Japan.
Thus far, the main focus among academics has been developing new analytical tools and methodologies for efficiently processing and making sense of massive volumes of real-time information and using these tools to better understand virtual social networks. Applied researchers, planners, and policy makers are only just beginning to explore the potential of Big Data to help clarify social attitudes and potentially inform local policy and development decisions.
As of yet, few have studied whether and how Twitter posts can be used to better understand peopleās perception of placeāthat is, how they actually feel about the communities in which they live, work, and play. 1 In the business world, firms have been doing this social media ālisteningā for years; they call it social listening (Hinchcliffe and Kim 2012; Rappaport 2011). In this book, we push forward a new research agenda that advances knowledge and methods around urban social listening.
Why does it matter what peopleās perceptions of places are? Among other reasons, urban social listening can help policy makers and planners understand the overall sentiments and well-being of people living in urban areas. This is critically important particularly because the redevelopment potential of many post-industrial cities has long been stymied by the negative perceptions of investors and potential new residents and businesses. Furthermore, strong place attachment can be a galvanizing force behind community renewal efforts, which are sustained through the dedication, will, and sweat of residents who care deeply about their communities.
Researchers have begun exploring, for example, how they might determine the subjective well-being (SWB, the term used as a stand-in for happiness in psychological literature) of individuals based upon Facebook status updates (Kim and Lee 2011); how ātweetsā sent by users of Twitter might be used in assisting with emergency preparedness efforts for natural disasters, epidemics, and social uprisings (Merchant et al. 2011); and how tweets provide valuable land use information for urban planners (Frias-Martinez and Frias-Martinez 2014).
In making use of the large quantities of microblogging data now available to researchers, this book will examine the sentiments residents have in places they occupy. Two central arguments underlie the book: (1) Just as social media has revolutionized social life, social listening can revolutionize the way that social scientists study cities and (2) that microblogging data is a rich data source with which to commence this new wave of urban research.
We will present in the following chapters the idea that, with the era of Big Data upon us, the study of cities will never be the same as it has been in the past. Attention to what millions of ordinary citizens are saying, within confined and narrow geographies, can provide more valid and reliable results than the obtrusive measures that have characterized social science research for more than a century. Likewise, microblogging data offers an abundant data source with which to do that listening.
While there have certainly been some naysayers (Goodspeed 2013) who question the validity of studies using these unobtrusive data sources, many others have adopted the new medium with aplomb. Some researchers have employed social network analysis to explore the ways in which individuals interact with one another (Hansen et al. 2009; Ediger et al. 2010; Catanese et al. 2010) or the ways in which people follow links (Namata et al. 2010), or combing content with user comments (Eugene Agichtein et al. 2008). Emoticons have also been analyzed to help consumers, marketers, and organizations use sentiment analysis to research products or services and analyze corresponding customer satisfaction (Go et al. 2009). Much of this research uses social media to understand group processes and properties (Tang and Lui 2010) but does little to fundamentally reveal what people think about places.
Important social science research has used massive social media data sets to advance social objectives (Ediger et al. 2010), to forecast shifts in the mood of users (Servi and Elson 2012), to enhance journalistic investigations (Diakopoulos and Shamma 2010), and to infer usersā locations from their tweets (Mahmud et al. 2012). For this book, we follow a tradition of using social media to conduct opinion mining and sentiment analysis (Gokulakrishnan et al. 2012; Martineau and Finin 2009; Meeyoung Cha et al. 2010).
Although many studies have been conducted using Big Data, including some that make an effort to gauge the emotional states of users, very little work has attempted to incorporate psychological theory to interpret exactly what form of emotional well-being microblogging data measures. Moreover, there are many differing methodologies that have been used to deciphe...