Loading...
 

Using Data Visualization to Analyze Big Data in Social Networks

Contributor: Tracey Hayes
Affiliation: Gonzaga University
Email: hayest at gonzaga.edu
Released: 15 January 2021
Published: Spring 2021 (Issue 25.2)

Introduction

Today social networks allow protests to develop using complex components and strategies; furthermore, new tools for digital analysis allow scholars to study patterns and connections in those social movements analyzing online protests and the complex rhetorical work and connections occurring within an online protest (Hayes, 2016). The tools and programs available to study social media in many ways make the process easier, in regards to the amount and type of data available. Nonetheless, this increase in available data presents challenges as data must be collected, sorted, selected, and analyzed. The options present many difficult choices as much of this is charting new territory, along with new methods.

When using digital methods, Richard Rogers (2013) advocates the understanding and usage of not only the digital devices, but how the digital objects created by these devices can assist in creating research questions. He asks how these digital objects such as hashtags, followers, tweets or locations can be juxtaposed and used as more than a means for finding information, but for using information. Using these digital objects does require tools, as Derek Hansen, Ben Shneiderman, & Marc A. Smith (2011) discuss the advantages of using social network analysis (SNA), “Social network analysis offers a systematic method to evaluate social media efforts, replacing anecdotes with scientifically based evidence” (p.8). Their discussion focuses on the governmental and business uses of SNA to determine the successes and failures of social media strategies, these same concepts can apply to understanding online protests or online social movements. Additionally, Jeremy Foote, Aaron Shaw & Benjamin Mako Hill (2017) explain the ability of network analysis to take large quantities of data to explore relationships and connections.

Twitter is showcased as an entity that provides digital objects to examine relationships and connections as well as the success of a hashtag campaign. Axel Bruns, Tim Highfield & Jean Burgess (2013) study two separate online protests during the Arab Spring in order to analyze the connections and language (Arabic and English) between the two campaigns using the hashtags #egypt and #libya. They discovered the use of Twitter was complex and while Twitter played a role in the quest for regime change, it was not the only method or tool used. Axel Bruns (2012) used SNA to examine conversations occurring on Twitter through analyzing the use of hashtags along with @replies and retweets. This analysis enabled the visualization of how participants participated, the roles they played, and how conversations evolved over time. Additionally, Axel Bruns, Jean Burgess, & Tim Highfield (2014) gathered multiple hashtags covering different Australian events to understand how Australians overall are using Twitter versus focusing on one event/one hashtag.

Online protests consisting of hashtag activism have focused on a multitude of issues ranging from economic, sexual assault, racial equality, police brutality, and social justice. Some examples include #BlackLivesMatter, #OccupyWallstreet, #BringBackOurGirls, and #MeToo, along with the recent #GetMePPE, which is a call to showcase the lack of Personal Protective Equipment that medical workers and first responders are encountering in their work against COVID-19.

Using digital tools to collect, display, and analyze data using social network analysis can assist in developing and answering research questions within these (and other) online social movements/protests to understand how participants create connections, construct knowledge, and share through their digital discourse and writing. This webtext will explain how social network analysis (SNA) and other data visualizations can assist digital humanists in researching and answering questions in analyzing big data related to protest movements through a case study of the SNA used in analyzing #MyNYPD.

Case Study

The online protest in this case study examined through SNA is the #MyNYPD protest, which occurred on April 22, 2014 when The New York City Police Department (NYPD) tweeted out a request (see Figure 1), “Do you have a photo w/ a member of the NYPD? Tweet us & tag it #MyNYPD. It may be featured on our Facebook page.” The NYPD quickly received responses from the public, but instead of positive photos with the police, the public organized online by repurposing the hashtag to share images of police abuse and brutality in an effort to subvert a public relations campaign into an online protest (Hayes, 2017, p. 120). Despite the often-short-lived nature of Twitter protests, this protest continued in the days, weeks, and months that followed. The data (tweets containing the hashtag #MyNYPD) collected over a time period of six months from January 13, 2015 – July 12, 2015 contained tweets from several months to over a year after the initial incident, thus providing an example of an extended protest.

This topic has gained relevance with the 2016 Presidential election and the increase in activism as demonstrated by the number of protests after Trump assumed office (e.g., Women’s March – January 21, 2017, Airport Protests against the Travel Ban – January 28, 2017, and March for Our Lives – March 24, 2018). Thus, SNA can assist in answering questions related to engagement and participation, in order for online protests to mobilize participants and sustain interest in their causes over the long term.

Man smiling and posing with two NYPD officers.
Fig. 1. Initial tweet from the New York Police Department. The tweet reads, “Do you have a photo w/ a member of the NYPD? Tweet us & tag it #myNYPD.”


Specifically, this case study consisted of a social network analysis of the five days containing the most tweets from the six months of collected tweets in order to understand the connections and communities formed within the #MyNYPD protest. Through examining the most proficient tweeters, either by the number of tweets or the number of retweets their tweets garnered, the connections between different players within the protest were discovered. Examining those connections helped to understand how tweets were used to mobilize, engage, and sustain interest in an online protest.  

Data containing the hashtag #MyNYPD was collected on a weekly basis for a six-month period (January 13, 2015 – July 12, 2015) using NodeXL ("Social Media Research Foundation," n.d.) and Gephi ("Gephi.org," n.d.) was used to graph the data. These graphs were then used to answer the following research questions.

  1. What are the relationships between and among users’ activity online?
  2. How are people connecting via Twitter within this protest?

To answer these questions, the five days with the highest number of interactions were selected to conduct a social network analysis for each of the five days. Interactions can consist of tweets (with or without engagement from another user) or retweets, mentions, or replies to (which are all considered engagement). The five days with the most interactions were selected in order to offer the broadest number of interactions/engagements to analyze the connections between participants within the online protest.

Since the focus of the case study is Tweets (Twitter data), it is important to acknowledge that using Twitter as a data source requires forethought in both data collection and analysis. Regardless of the application chosen to collect Twitter data, there are issues with the retrieval of historical data, primarily involving expense or advanced technical knowledge. Therefore, if a particular hashtag appears of interest, collecting the data immediately is important, as most software applications only provide access to historical tweets from 7-9 days ago. Sometimes there are Twitter datasets, generally for the more popular hashtags, collected and made available for researchers to share/download. See https://catalog.docnow.io/ ("DocNow Catalog," n.d.).

While this webtext focuses on SNA as data visualizations, there are other visualizations in which the same data used for an SNA can be presented. One example is the visualization of hashtag usage.  The top ten hashtags (Figure 2) used in combination with the primary hashtag of #MyNYPD (Hayes, 2017) show the topics and connections within the #MyNYPD online protest. Examining these secondary hashtags provided insight and details into how (in this case) a large-scale protest against the New York Police Department was connected to specific incidents, other protests, or other police departments, thus visualizing connections between the primary hashtag and secondary hashtags used in a tweet.

Graphic showing the Top 10 secondary hashtags (FTP, myELAS, NYPD, NYC, myLAPD, EricGarner, Justice4Cecily, S17, fail, OWS) used with #MyNYPD
Fig. 2. Top 10 secondary hashtags related to the primary hashtag - #MyNYPD from Hashtagify.

 

Tools and Processes

The tools (including new computational computer software programs such as NodeXL and Gephi) available to study social media in many ways make the analysis easier, in regards to managing the amount and type of data available. Nonetheless, this increase in available data presents challenges as data must be collected, sorted, selected, and analyzed. Ethical and moral decisions may also be involved in the choices. These issues are not new as danah boyd and Kate Crawford (2011) discuss big data and the collection, analyzing, and using of big data. The call continues today as William L. Wolff (2017) urges Twitter scholars to explain the methods, processes and decisions used in collecting data, using data, and displaying data, both practical and ethical. However, in order to conduct social media research in an ethical manner, researchers have to start at the beginning and understand how the software that facilitates this research works and is utilized; Bawden (2008) calls it the “underpinnings” in relation to computer literacy. Understanding how software such as NodeXL works is foundational to collecting data (the first step), with the second step requiring an understanding of advanced software used to analyze big data within social media research. The following section illustrates and explains how NodeXL and Gephi work to create graphs as a part of SNA.

Node XL

NodeXL is a Microsoft Excel template that allows for importing data from a multitude of social media or network applications, including Twitter. NodeXL also allows for data to be exported into different social network analysis software (e.g., GraphML, Pajeck). While NodeXL does offer a free Basic version, one of the drawbacks is the more robust version is a paid version. NodeXL also offers examples of social networks created with NodeXL along with details about the data found through these graphs. For example, the graph 4Cinsights provides information regarding the number of tweets, dates of data collection, and algorithms used along with the data found within these tweets such as the top influencers, top domains, top hashtags, top URLs. “NodeXL Basic and NodeXL Pro are add-ins for Microsoft® Excel® (2007, 2010, 2013, 2016) that support social network and content analysis” as explained on the NodeXL website ("Social Media Research Foundation," n.d.). Basic NodeXL provides users the opportunity to use and explore the software and make a decision if NodeXL is useful or if they would like to upgrade to NodeXL Pro. NodeXL Pro can import data from Twitter, Facebook, YouTube, Flickr, and MediaWiki.

Gephi

Gephi is open-source software that allows for network visualization and analysis of data. As Gephi states, “Gephi is a tool for data analysts and scientists keen to explore and understand graphs. Like Photoshop™ but for graph data, the user interacts with the representation, manipulates the structures, shapes and colors to reveal hidden patterns. The goal is to help data analysts to make hypothesis, intuitively discover patterns, and isolate structure singularities or faults during data sourcing.” ("Gephi," n.d.). As such, Gephi assists in visualization data to see, understand, explore, and analyze connections within the data with additional options and flexibility beyond NodeXL.

The decision to use NodeXL to collect the data and Gephi to display the data at the time was based on the options Gephi provided to create the graphs and display the data, which Wolff (2017) acknowledges as a difficulty as well with NodeXL (its graphing capabilities). That being said, NodeXL keeps innovating and adding features, and depending on your goals, NodeXL may be all that is needed. One example is that NodeXL now includes the number of retweets or favorites for a given interaction. Additionally, there are differences in visualizing historical tweets versus tools that create real-time visualization (see Gephi’s plugin - Twitter Streaming Importer).

Force Atlas

Force Atlas is an algorithm used to distribute nodes and edges spatially within a network graph created by Mathieu Jacomy. Without an algorithm, the data is difficult to analyze since it is clustered together. The different algorithms produce different spatial arrangements. Force Atlas was chosen because, “It is scaled for small to medium-size graphs, and is adapted to qualitative interpretation of graphs” (Jacomy, 2011).

Data Analysis Process

First, the data for each day was exported into a GraphML format from NodeXL that could be imported into Gephi for analysis. Using Gephi, Force Atlas was applied to create a visualization that could be easily read (when data is first imported, the data is jumbled together), thus allowing for an analysis of the particular day’s data including the who, what, where, when, and why of the #MyNYPD sample for that day. Force Atlas displays the nodes and edges based on their connections. The repulsion strength attribute was set to 10,000, which separates the different groups of interactions (“communities”) spatially in the graph and illustrates that the greater the repulsion strength, the greater the distance between nodes that are different from each other. If the repulsion strength were less, the communities would be closer. The term communities refers to nodes that connect to edges in a grouping; however, the word is not used in the sense of a community gathering together in an affinity space, because this is not necessarily the case in each group of interactions. Rather, “community” here denotes users who have interactions with each other’s tweets. Please note that SNA is flexible; different graphs can be created with the same data using different algorithms in order to answer different research questions.

Changing Technology Notes

It is important to note that since this analysis, the technology and methods may have changed since the data was collected in 2015 and analyzed in 2015/2016. Rapid changes in technology present another reason for sharing knowledge and resources as building blocks to understanding the differences and having the ability to adjust to these changes, although many times changes in technology make tasks less cumbersome and more user-friendly, but this is not always the case.

Understanding Network Analysis: Nodes and Edges

The network used for this analysis will be from June 5, 2015 (See Figure 3).

Social Network Graph of #MyNYPD from June 5, 2015
Fig. 3. Social Network Graph of #MyNYPD from June 5, 2015

 

General Overview of June 5, 2015

A key finding from this analysis is that multiple people can participate in the protest on the same day with different goals. Figure 3 contains 130 nodes and 130 edges, indicating (and is proven after closer inspection) that each node has an interaction with another node, and only that node. There are not nodes interacting with other nodes in their communities on this particular day. Also, since we know that there are 319 interactions in this data set, it appears that one or several people are responsible for multiple interactions on June 5, 2015.

In contrast to Figure 3, Figure 4 shows an example of different nodes interacting with each other. There are a total of five nodes and seven edges. Both Node A and Node B have multiple interactions with multiple nodes, hence their larger sizes. Node A retweeted/mentioned three nodes, while Node B retweeted/mentioned four Nodes including Node A.  

Graphic of nodes interacting with multiple nodes
Fig. 4. Nodes interacting with multiple nodes.


An initial inspection of Figure 3 shows there are two major players on this day, as well as a few smaller communities, and several orphans. The two main players are GracieeGorgeous with 96 interactions, and Copwatch with 23 interactions. Figure 5 provides a closer look at Copwatch's interactions and Figure 12 provides a closer look at GracieeGorgeous and her interactions.

Components of a Network

Before interpretation of a network can begin, understanding the different components that comprise a network and how they interact is important. The basic components of a network are nodes and edges; the nodes are represented by circles and the edges by the paths that connect the nodes. Within this case study, the nodes are Twitter users and the edges represent the interactions between the users. Additionally, there are two types of networks, directed and undirected. In a directed network the edges have an arrow at the end of the edge to indicate which node created the interaction and is thus the source of the interaction, undirected networks do not have an indication of who is the source, the edges only indicate a relationship between the two nodes.

The differences in participation are noted by the different sizes and colors of the nodes. In Figure 5 below the community member Copwatch is the largest node, and is a different color (beige/gold) thus indicating the highest level of participation within this community. Within this case study, the term community refers to participants (Twitter users) with an interaction(s) (tweets/retweet/mentions/replies) within the specific network.

Copwatch’s Network

Social network graph of Twitter user Copwatch's interactions on June 5, 2015
Fig. 5. Graphic of Twitter user Copwatch’s interactions on June 5, 2015.

 
Copwatch’s tweet that created this network can be seen in Figure 6. The network shows the 23 interactions Copwatch’s tweet created. There can be a variance in number of interactions and likes based upon when the tweet is viewed versus when the data is collected by NodeXL unless the two are collected at the same time. For example, the network above shows 23 interactions, while the tweet itself (see Figure 6) shows 28 retweets. The 23 interactions from Copwatch’s tweet include 22 retweets and the tweet itself is counted as one interaction.

Image of Copwatch's tweet on June 5, 2015
Fig. 6. Copwatch’s tweet on June 5, 2015.

 
Zooming in on Copwatch’s interactions (See Figure 7), the edges all have arrows directed towards Copwatch, meaning that Copwatch is the source of the interactions. The edges are also the same beige/gold color that Copwatch’s node is, thus separating these edges from other edges within the entire days’ interactions. The color choice of the edges was selected in Gephi (See Figure 8), and the decision was made to have the edges be the same color as the source, as the arrows can at times be difficult to view in a clustered community. Therefore the edge color can serve the same purpose as the directional arrow in a directed network.

Close-up of Copwatch’s graphed interactions from June 5, 2015.
Fig. 7. Close-up of Copwatch’s graphed interactions from June 5, 2015.

 

Graphic showing the settings for Edge Color in Gephi
Fig. 8. Settings for edge color in Gephi.

 

As the second most active community member on June 5, 2015 with 23 interactions (1 tweet and 22 retweets), Copwatch used a different approach. Copwatch only tweeted once, and the tweet was retweeted 22 times. There is not any use of other Twitter user names, as Copwatch is not attempting to connect on an individual level, but on a more global level sending out one tweet, and hoping that followers will retweet and spread the message. Similar to GracieeGorgeous, there are not any interactions between other nodes in the community.

suziq2opn’s Network

However, there is another active participant in this community, participating in a different way than GracieeGorgeous or Copwatch, and that is suziq2opn. suziq2opn has three interactions (much less than the other major players), but her interactions consist of three different activities (see Figure 9). suziq2opn is set apart from the rest of the nodes interacting with Copwatch because suziq2opn has additional interactions other than only retweeting Copwatch, which is her first interaction. Her second interaction is a tweet without any interactions. This is visualized via an edge that starts with suziq2opn and connects to suziq2opn (the edge is a small red half circle on the right side of the suziq2opn’s node). This edge visualizes suziq2opn as the source and target, which is important especially when there are not any interactions, the tweet is still visualized (although there are settings within Gephi to set the minimum number of interactions to visualize within the graph). The tweet that created this second interaction can be viewed in Figure 10. suziq2opn has a third interaction created from an additional tweet in which suziq2opn mentions blackvoices. This interaction creates an edge with an arrow directed towards blackvoices, as blackvoices is the target of the interaction. (See Figure 11 for the tweet that created this third interaction.)

Graphic showing suziq2opn’s interactions from June 5, 2015
Fig. 9. Graphic showing suziq2opn’s interactions from June 5, 2015.
 
suziq2opn’s Tweet from June 15, 2015.
Fig. 10. suziq2opn’s Tweet from June 15, 2015.

 

Screeenshot of suziq2opn’s tweet mentioning @blackvoices
Fig. 11. suziq2opn’s tweet mentioning @blackvoices. 


Some users change their Twitter username. This can be seen with the tweets from suziq2opn, which was the username when the data was collected, but as seen in Figures 10 and 11, suziq2opn’s username has changed to suzi_q2.

NodeXL uses UTC, which stands for Coordinated Universal Time. Therefore, the date/time on tweets will be different than the NodeXL data. For example, in Figure 10 the time on the tweet is 2:16pm, while in NodeXL the time for the tweet is 21:16 (9:16pm). Therefore, a day’s data for a 24-hour time period varies and this might have an impact with data visualization and analysis. UTC does not have daylight savings time either.

GracieGorgeous’s Network

The major community member in this network is GracieGorgeous (See Figure 12).

Graphic of GracieGorgeous interactions on June 5, 201
Fig. 12. Graphic of GracieGorgeous interactions on June 5, 2015.


While it may appear at first glance that the interactions between GracieGorgeous and Copwatch are similar, the exact opposite is occurring. Copwatch tweeted and then was retweeted 22 times, while GracieGorgeous’s interactions are created through multiple tweets, mentioning multiple usernames (see Figures 13 and 14), hence the edges are red, the color of the target nodes, as well as directed towards these nodes.

Screenshot of GracieGorgeous's tweet on June 5, 2015
Fig. 13. GracieGorgeous tweet from June 5, 2015.

 

Screenshot of another GracieGorgeous's tweet on June 5, 2015
Fig. 14. Another GracieGorgeous tweet from June 5, 2015.


In this example, GracieeGorgeous attempted to disseminate information through multiple tweets. Visually we can see that GracieeGorgeous is the main player in this community, as she circulated information regarding her personal experience with the police, and the Manhattan District Attorney. This tweet (see Figure 13) has four interactions as she mentioned @newyorkcity (New York City information), @Manhattanda (the Manhattan District Attorney Cyrus Vance), @SafeHorizon (Victims organization), and @TheJHF (Support for survivors of abuse and violence), along with a link to a Tumbler page with more details about what happened to her.

The tweets that followed this one were very similar, with slight variations depending on usernames she was mentioning and the character length of said usernames (see Figure 14). Her multiple tweets addressed people as diverse as politicians (Hillary Clinton and Bill de Blasio), support groups for victims (Safe Horizon and Joyful Heart Foundation), other affiliated New York City Police Department (NYPD Special Victims Unit, NYPD Special Ops). She also appeared to be contacting journalists, but the usernames she tweeted for Mariane Pearl (@mariannepearl) and Maureen Dowd (@maureendowd1) are not the journalists’ Twitter accounts. Mariane Pearl’s Twitter account is @MarianePearl (notice there is only one n) and Maureen Dowd is @NYTimesDowd. It appears that GracieeGorgeous entered Twitter user accounts using actual names, but in several cases, those were not who she appeared to be trying to contact. She guessed right on Clinton and de Blasio, but not on the two journalists. While her tweets varied in whom they were directed to, the message was the same; she included @newyorkcity and @manhattanda in these tweets, hence the rather large red arrows pointing towards @newyorkcity and @manhattanda. The arrows are larger to represent the number of times GracieeGorgeous mentioned those two users in her tweets. All the other edges consist of a lighter weight and smaller arrows indicating one interaction, except for @voicelessnomore, who did retweet GracieeGorgeous’ tweet, hence the weighted edge.

GracieeGorgeous’s tweets present a different type of activism within this protest, as she takes advantage of the affordances Twitter offers. GracieeGorgeous employs a shotgun approach, sending out multiple tweets to multiple people in an attempt to reach as many people as possible. Business Dictionary.com defines the shotgun approach as, “Marketing strategy whereby (in contrast to rifle approach) the aim is to cover as wide an area or population as possible.” Certainly, the more people she notifies, the better chance somebody will listen. Of course, it can be argued that she did not receive many retweets, and therefore was unsuccessful in her attempts; however, she mostly targeted high profile people, as well as the opposition. Chances of getting responses from Hillary Clinton or Bill de Blasio were small, harkens back to the definition of successful activism. GracieeGorgeous participated in the protest, in this instance on a personal level regarding her encounters with the NYPD, and her participation rate was high.  Although her strategy from this vantage point may not have been as successful, she still contributed to the protest, and was a participant in this online protest. Additionally, while her credibility may have taken a hit due to her “guessing” of Twitter usernames, it appears to be a common occurrence. Of course, more research would be needed, but a search for @MaureenDowd on Twitter shows that multiple people are committing the same mistake. Additionally, at this time the account @MaureenDowd has been suspended, perhaps that is why GracieeGorgeous used @MaureenDowd1.

Conclusions

While GracieeGorgeous and Copwatch both are circulating information regarding the NYPD’s abuse of power, Copwatch concentrates on information regarding another person and their police encounters, while GracieeGorgeous concentrates on an issue pertaining to her. Perhaps that is why she tweeted so many times to so many people; she wanted others to know what had happened to her. Additionally, Copwatch is a Twitter account that promotes filming and taking photos of the police, so while not a corporation per se, Copwatch is not an individual person either. suziq2opn may not have as many interactions as either of the two main players, but she is just as vital a participant as she tried to engage multiple nodes by retweeting and mentioning. She also expresses her own viewpoint through a tweet, so suziq2opn is not only a participant but a creator as well. Her two interactions with other nodes though are not individuals, but entities. GracieeGorgeous, Copwatch, and suziq2opn all participated in this protest using different methods and different purposes, showing that there is more than one way to tweet even though they are all participating in the same affinity space, thus showing the interactions and communities within this protest as multi-dimensional through their diversity and complexity.

SNA is an integral part of digital humanities research in a wide variety of disciplines. For example, the Six Degrees of Francis Bacon project illustrates a data visualization of the social network of Francis Bacon. The creators of this project (faculty, staff, and students at Carnegie Mellon) have made the code open source and data available on Github for others to analyze this data and utilize it in their own projects. Additionally, writing this article has caused me to advocate strongly for articles, wikis, videos, and tutorials that are up-to-date and can be used by those without advanced programming or technology knowledge. As Luca Hammer (2016) states in his guide, Analyzing Twitter Networks with Gephi 0.9.1, “Get ready to get your hands dirty and spend hours without getting anywhere and keep going or starting fresh the next day,” thus showing the exigency for reference materials and support communities where users can seek and receive up-to-date advice, tips, and answers.

References

Bawden, David. (2008). Origins and concepts of digital literacies in Collin Lankshear & Michele Knobel (Eds.), Digital literacies (pp.17-32). New York, NY: Peter Lang.

Bruns, Axel. (2012). How long is a tweet? Mapping dynamic conversation networks on Twitter using Gawk and Gephi, Information. Communication & Society, 15(9), 1323-1351. https://10.1080/1369118X.2011.635214

Bruns, Axel, Burgess, Jean, & Highfield, Tim. (2014). A ‘big data’ approach to mapping the Australian Twittersphere. In Paul Longley Arthur & Katherine Bode (Eds.), Advancing Digital Humanities (pp.113-129). Palgrave Macmillan, London.

Bruns, Axel, Highfield, Tim, & Burgess, Jean. (2013). The Arab Spring and social media audiences: English and Arabic twitter users and their networks. American Behavioral Scientist, 57(7), 871–898. https://doi.org/10.1177/0002764213479374

boyd, danah & Crawford, Kate. (2011, September 21). Six provocations for big data. A decade in Internet Time: Symposium on the Dynamics of the Internet and Society. http://dx.doi.org/10.2139/ssrn.1926431

Copwatch. (2015, Jun 5). Teenage girl wins $45,000 settlement after spending night in jail for rejecting NYPD cop’s flirty advances http://photographyisnotacrime.com/2015/06/undercover-nypd-officer-arrests-17-year-old-girl-after-she-rejects-his-flirty-gestures-resulting-in-settlement/ … #myNYPD Tweet. Retrieved April 2, 2018 from https://twitter.com/copwatch/status/606928919997116416

DocNow Catalog. (n.d.) Tweet ID datasets. Retrieved January 16, 2019, from https://catalog.docnow.io/

Foote, Jeremy, Shaw, Aaron & Hill, Benjamin Mako. (2017). A computational analysis of social media scholarship. In Jean Burgess, Alice Marwick, & Thomas Poell (Eds.), The SAGE Handbook of Social Media (pp. 111-134). Los Angeles, CA: SAGE Reference.

Gephi.org (2016). Gephi (Version 0.9.1). Gephi.org. https://gephi.org/

GracieeGorgeous. (2015, Jun 05). @SBANYPD Women of @newyorkcity BEWARE U aren't SAFE when U turn 2 #MyNYPD & @ManhattanDA 4 help w/#domesticabuse Tweet. Retrieved April 2, 2018, from https://twitter.com/gracieegorgeous/status/606901800709570560

GracieeGorgeous. (2015, Jun 05). Women of @newyorkcity BEWARE U arent SAFE when U turn 2 #MyNYPD & @Manhattanda 4 help w/ABUSE http://t.co/mdwlupARcx @SafeHorizon @TheJHF Tweet. Retrieved April 2, 2018, from https://twitter.com/gracieegorgeous/status/606890056951939072

Hansen, Derek. L., Schneiderman, Ben, & Smith, Marc A. (2011). Analyzing social media networks with NodeXL: Insights from a connected world. Burlington, MA: Morgan Kaufmann.

Hammer, Luca (2016, September 6) Guide: Analyzing Twitter networks with Gephi 0.9.1. Retrieved March 12, 2018, from https://lucahammer.com/2016/09/06/guide-analyzing-twitter-networks-with-gephi-0-9-1/

Hayes, Tracey J. (2016). What counts as successful online activism: The case of #MyNYPD. Doctoral dissertation, Arizona state University. ASU Library Digital Repository.

Hayes, Tracey J. (2017). #MyNYPD: Transforming Twitter into a public place for protest. Computers and Composition, 43, 118–134. https://doi.org/10.1016/j.compcom.2016.11.003

Jacomy, Mathieu. (2011, June 6). ForceAtlas2, the new version of our home-brew layout. Gephi Blog. https://gephi.wordpress.com/2011/06/06/forceatlas2-the-new-version-of-our-home-brew-layout/

NYPDnews. (2014, Apr 22). Do you have a photo w/ a member of the NYPD? Tweet us & tag it #myNYPD. It may be featured on our Facebook. http://t.co/mE2c3oSmm6 Tweet. Retrieved April 24, 2016 from https://twitter.com/NYPDnews/status/458665477409996800

Rogers, Richard. (2013). Digital methods. Cambridge, Massachusetts: The MIT Press.

Shotgun Approach. (n.d.). In BusinessDictionary.com. Retrieved from http://www.businessdictionary.com/definition/shotgun-approach.html

Social Media Research Foundation. (2019). NodeXL (NodeXL Pro). Social Media Research Foundation. https://www.smrfoundation.org/

suzi_q2. (2015, Jun 5). Mother Of Man Who Died In Police Custody Sues, Hoping For Release Of Arrest Video http://huff.to/1KKqfQS @blackvoices #myNYPD #DeathByCop Tweet. Retrieved April 2, 2018, from https://twitter.com/suzi_q2/status/606961478072016896

suzi_q2. (2015, Jun 5). #myNYPD #DIZZNELAND #NYC Tweet. Retrieved April 2, 2018, from https://twitter.com/suzi_q2/status/606932747601342464

Wolff, William L. (2017, October 1). Twitter archives: A discussion of systems, methods, visualizations, and ethics. Kairos: A Journal of Rhetoric, Technology, and Pedagogy, 22.2. Retrieved March 15, 2018 from http://praxis.technorhetoric.net/tiki-index.php?page=PraxisWiki%3A_%3ATwitter+Archives


Created by kristi. Last Modification: Wednesday January 13, 2021 22:22:16 GMT-0000 by kristi.