What Are We Doing? A Study of Collaborative Storytelling Through Twitter

Embed Size (px)

Citation preview

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    1/15

    Jay ChangRachel Lerner

    Megan McGrathMichael Nutt

    Kelly Smith

    WHAT ARE WEDOING?A Study of Collaborative Storytelling Through Twitter

    I. Introduction

    In recent years, we have seen a growth in the participation of online collaborative

    storytelling on social network websites such as Twitter. For our study, we definestorytelling as the online, interactive, public sharing of events. Social networking sites

    (SNSs) sometimes provide a platform for storytelling by allowing users to publiclybroadcast personal experiences and engage in dialog with others in real time. These kinds

    of interactions provide a rich contextual lens for the examination of events. Dialog froman SNS can be woven into other accounts of an event to form the fabric of a story. This

    paper reviews the literature surrounding this form of storytelling and proposes a study tobe conducted at the University of North Carolina at Chapel Hill. The study examines

    UNC-CH affiliates' use of the SNS Twitter.com to tweet, or publicly share text-basedinformation, about UNC-CH related events. In particular, we will be looking at whether

    the Twitter network views information workers as authoritative sources of informationabout events. We define information workers as faculty, staff, and students associated

    with campus libraries and the School of Information and Library Science. The purposeof our research is to move towards a fuller understanding of the users that engage in

    storytelling via Twitter and the stories they tell.

    Current literature provides us with the background to set up our study. First, welook at previous research concerned with the act of microblogging. The literatureprovides information on how the act of microblogging was created and how it is used in

    various settings. Additionally, while microblogging and blogging are related, there areseveral distinctions, especially those related to time and frequency. Next, we look at

    works on informal communication. As an online form of text-based contact, tweets areclassified as a mode of informal communication. Zhao & Rosson (2009) provide an

    explanation of the two main benefits of informal communication - personal and relational- and its effectiveness in communicating to a large target audience. Literature on using

    online tools and Web 2.0 applications for information sharing is relevant to our study.Previous studies, which have analyzed Twitters use as a marketing tool, have described a

    phenomenon called electronic word-of-mouth communication, which is an importantaspect to our study of Twitter. In addition, several works discuss various motivations

    behind using Twitter. Literature on the tool Twitter itself provides us with an overviewof the site, its community of users, and its various uses. In addition, one study discusses

    the geographic elements of communities and how they may be used to link to events.Finally we examine the literature on storytelling in general so that we could apply it to an

    online collaborative environment.

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    2/15

    In our study of online collaborative storytelling using communication throughTwitter about UNC related events, our efforts are guided by the following question: Does

    greater authority on Twitter equal a greater authority in the historical record? While ourstudy cannot answer this question, we hope to provide data that will allow us to think

    more critically about the question. Our study is grounded in two hypotheses: (1) Events

    related to information workers are more likely to be tweeted about than other events, and(2) Information workers are more likely to be valued sources of information about eventson campus than other Twitter users. To conduct our study, we will monitor tweets with

    the online data harvesting tool ContextMiner, developed at the UNC-CH School ofInformation and Library Science by Chirag Shah.

    II. Literature Review

    2.1 Introduction

    Microblogging, the communication mechanic used on Twitter, is a relatively new

    phenomenon. Although literature defining what it is, how it works, and its practicalapplications is still being produced, it seems prudent to look at how microblogging cameto be, who microblogs, why people use microblogs, and how microblogging is used in a

    variety of settings. Although many of our sources do not explicitly analyze Twitter, ourfocus in this study, the conclusions drawn from general microblogging research are still

    applicable.

    2.2 What is Microblogging?

    Microblogging can be defined as short, character-restricted remarks distributed toa predefined network, with the option for wider dispersal or re-distribution (Jansen,

    Zhang, Sobel, & Chowdury, 2009). It is defined more specifically by Erikson (2010) as aterm used to describe social networking actions which involve "the textual authoring and

    posting of 140 character missives, sometimes called status updates. Twitters implicitprompt to users is to answer the question What are you doing? (Erickson, 2010).

    There are numerous websites through which one can produce a microblog, and evenmore methods for wide distribution of the content. For the purposes of this study, we are

    applying this research to the specific microblogging platform Twitter.Though it is recognized that microblogging is related to blogging, there are some

    key distinctions. Java, Song, Finin, and Tseng (2007) describe the distinctions as two-

    fold: time and frequency. While microblogs are generally updated several times a day,blogs in general are updated only once a day. Microblogging requires less commitment of

    time and thought to create content through the use of character-restricted posts, and thusfulfills a need for an even faster mode of communication (Java et al., 2007, p. 57).

    Though the usage patterns differ for each, it has been shown that motivations forblog and microblog use include storytelling, information sharing, and opinion sharing. In

    an interesting study regarding the motivations of bloggers, Nardi, Schiano, Gumbrecht,and Swartz (2004) found that blogs are used as a medium for sharing experiences and

    opinions.

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    3/15

    Similarly, Java et al. (2007) found that the main types of user intentions in Twitterare: daily chatter, conversations, sharing information, and reporting news. Motivations

    for microblogging will be addressed specifically in another section. However, it isimportant to note similarities among online text-based communication tools and how

    each new tool or method is born of the previous iteration.

    2.3 Informal Communication

    Online communication forums such as microblogs are an example of informalcommunication - the equivalent of water cooler conversations at work (Zhao & Rosson,

    2009). Zhao and Rosson provide a very clear chart (Fig. 1) on the two main benefits ofinformal communication.

    Figure 1: Benfits of Informal Communication

    (Zhao & Rosson, 2009)

    Informal communication leads to better management of personal relationships, as

    it allows one to more accurately assess the abilities, personalities, characters, andbehaviors of others (Zhao & Rosson, 2009). By engaging in microbloggingconversations, one is allowed access to what others are doing, to whom they are talking,

    and what their interests are. These are all means by which one normally assesses apersons character; through informal electronic communication, however, one can do this

    before meeting a person face-to-face. Additionally, these quick, informal, onlineinteractions can herald feelings of closeness, collectivism, and connectedness between

    otherwise informal relationships, such as among colleagues (Nardi, 2005). Personalbenefits of informal communication are seen through the acquisition of new information

    that relates to ones personal information needs (Zhao & Rosson, 2009).It seems counter-intuitive, but people who study social networks suggest that

    more novel - interesting, useful, and productive - information is passed between weak ties(informal relationships) as opposed to strong ties (formal relationships). This is due to

    the fact that weak-tie relationships exist outside of our normal activities and social sphereand therefore provide information from a different perspective (Granovetter, 1973).

    Concurrently, feelings of mutual trust are more likely and are more quickly formedbetween weak ties (Levin, Cross, & Abrams, 2002). Zhao and Rosson conclude that,

    Weak ties are the most common social relationships that people have in their personalsocial networks. Thus an increased level of informal communication throughoutones

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    4/15

    social networkmay increase the chances of sharing and gaining valuable information(Zhao & Rosson, 2009, p. 244). In short, online informal communication, such as

    tweeting news of events on a campus, is an effective means by which to broadcast currentevents to a large, targeted net of people. When enough people use this form of

    communication about the same event, a larger story of the event emerges.

    2.4 Information Sharing

    Information sharing is an expansive concept in this study we are discussingsharing information collaboratively via online tools (specifically Twitter). Quite a bit has

    been written about web 2.0 tools, and how they have been used for collaboration,marketing, and storytelling within the workplace (Hastings, 2009; Zhao & Rosson, 2009;

    Jansen et al., 2009). Most relevant to our research, though, is the idea of electronic word-of-mouth communication and its implications for collaborative storytelling. Jansen et al.

    (2009) conducted a notable study on the use of Twitter as a branding tool for companiesthat illustrated how people are becoming markedly more influenced by the web

    communications by and about companies. The authors describe this phenomenon aselectronic word-of-mouth communication (EWOM).

    Traditional word-of-mouth communication must first be addressed, as EWOMstems directly from it. Word-of-mouth communication is generally considered a

    powerful tool for marketing it helps to influence people and operates via physical socialnetworks and trust (Jansen et al., 2009). In a similar manner, EWOM helps to influence

    people, but rather than operating via physical social networks and face-to-face (ortelephone) contact, it operates in the virtual world through online social networks, online

    reviews, and other forms of digital communication (Jansen et al., 2009). In fact, it hasbeen postulated that EWOM can be an even more powerful tool than word-of-mouth

    communication because it is can occur instantaneously, it has enormous reach, it isperceived as more trustworthy due to the fact that it is in print, and it is widely and easily

    accessible (Hennig-Thurau, Gwinner, Walsh, & Gremle, 2004).This relates to the literature described previously discussing weak-tie information

    sharing (Zhao & Rosson, 2009). If more novel, and therefore more desirable informationis acquired through weak ties, then EWOM is all the more effective, as more weak ties

    can be reached through EWOM than through face-to-face word-of-mouthcommunication. EWOM is also powerful due to the difficulty of stopping a word-of-

    mouth campaign (intentional or non-intentional) once it has gained momentum (Ennew,Banerjee, & Li, 2000). This, of course, becomes even more difficult to control in the

    virtual world, as items can be posted and reposted by hundreds of people in mereminutes.

    Microblogging, and Twitter in particular, blurs the lines between masscommunication and interpersonal communication. Jansen et al. sum it up nicely: These

    services provide constant connectivity among people that is previously unparalleled(Jansen et al., 2009, 2169). There are benefits to this philosophy of connectivity - a

    member of Zhao and Rossons survey of how Twitter and other microbloggingtechnologies are used in the workplace describes these tools as people-based RSS

    feeds. The idea is that one obtains useful information because one follows orsubscribes to people with interests similar to their own (Zhao & Rosson, 2009).

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    5/15

    The implications of being able to make connections and gain information fromsuch a large network are astounding. However, to make use of these technologies, one

    must first enroll, or buy into, the technology in the first place. In other words, in orderto create networks, share, and benefit from others sharing, one must first make an

    intentional participatory action one must sign up for the services (Cheung & Lee,

    2010). A barrier for entry for many people is having too much of their personalinformation online. Many users uncomfortable with Twitter's level of disclosure areoften confused or unaware of privacy boundaries or individual user settings (Palen &

    Dourish, 2003). Often, those who would benefit from using microblogs, blogs, and socialnetworking sites do not, for fear that their information will always be available to the

    general public.

    2.5 Motivation

    Attempting to determine the motivations of Twitter users, Java et al. (2007)analyzed why people tweet and established three main categories of users:Information

    Source,Friend, andInformation Seeker. AnInformation Source is a hub and has a largenumber of followers [and] may post updates on regular intervals or infrequently.

    Friend, the broadest category of the three, is comprised of friends, family, coworkers,and any other users who may expand their network of colleagues. AnInformation Seeker

    is a person who might post rarely, but follows other users regularly (p. 63). This studyorganizes users into three seemingly equal groups based on user intent.Friends, however,

    can also be viewed as the broader and overarching parent group of theInformationSource andInformation Seekeruser groups, as these sub-groups have more clearly

    defined user intentions. Reorganizing the relationships (Fig. 2) among user categoriesillustrates more clearly how an individualFrienduser may have multiple intentions or

    may even serve different roles in different communities (p. 63).

    Figure 2: Types of Twitter user groups.

    The varying motivations for using a microblogging site such as Twitter often

    overlap and apply to individual users in different ways. For example, users of Twitter ina business setting may seek different ends than casual users who simply use Twitter

    socially. Java et al. (2007) further conclude thatFriends, Information Seekers, andInformation Sources utilize Twitter in four main ways: 1)Daily Chatter, 2)

    Conversations, 3) Sharing Information/URLs, and 4)Reporting News. The largest

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    6/15

    category,Daily Chatter, consists of users' daily activities and answers to the questionposed by Twitter, What's happening? Conversations refer to the use of the @ symbol

    followed by a username to direct comments to a specific user or several users (e.g.@username). Sharing Information/URLs is sharing links to websites outside of Twitter.

    Finally,Reporting News consists of tweeting about current events, often including a link

    to the original news source. It is the fusion ofconversations and reporting news thatprimarily inform our definition of storytelling. Stories on SNSs consist not only ofreported news and events, but the ensuing conversations that follow shared news. The

    four user intentions listed above form what Cheung and Lee (2010) refer to as theCollective Intention or the We[Intention], where initial and continued use of Twitter

    relies on users successful interactions and their ability to build communities based onshared experiences.

    Communities form the basis of any social networking tool and are usually topicalor based on shared interests; communities and interests often overlap (Java et al., 2007,

    p. 60). These shared interests and intentions among Twitter users create common groundon which emerging communities are founded, where the value and reliability of

    information is of high importance. By choosing to "follow" another user, the follower isinvested in the user's broadcasts, and in turn "increases the perceived credibility of a

    subscribed information source" (Zhao & Rosson, 2009, p. 248). Forming communitiesallows people to have ambient intimacy, whereby users are able to keep in touch with

    people with a level of regularity and intimacy that [one] wouldnt usually have access to,because time and space conspire to make it impossible (Reichelt, 2007, para. 3).

    Conferences provide a real-life example. By following the tweets of a communitycovering a particular lecture, one can still gain insight into the material without being

    physically present. It is even possible to participate in concurrent events, using Twittercommunities to broadcast opinions and inquiries. Ultimately, events broadcast by

    information providers to information seekers strengthens the online community and addsto the success of physical events (Java et al., 2007).

    2.6 Twitter

    First appearing in 2006 as part of the Web 2.0 wave of social applications, Twitter

    is one of the most prominent social networking tools and is used by individuals as well ascompanies, news organizations, politicians, and other information providers who benefit

    from Twitter's simplicity and ease of use in broadcasting events (Java et al., 2007).According to its founders, Twitter "keeps you informed with what matters most to you

    today and helps you discover what might matter to you most tomorrow (Twitter, About,2010, para. 1).

    "What matters tomorrow" is important when analyzing Twitter through the lens ofonline collaborative storytelling of shared experiences. Twitter status updates form what

    Erickson (2010) calls citizen microbroadcasting, where information passed betweenindividuals is time sensitive; in others, the broadcasts act as color commentary on a

    special event (p. 1200). Once transmitted, the significance of tweeted events fades awayquickly, whereas non-event-related tweets may remain relevant for as long as they can be

    seen (Erickson, 2010).

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    7/15

    Twitter was originally established as a locationless social networking tool toconnect the world. However, the inclusion of locations both by users (in tweets about

    events) and by Twitter itself (which allows users the option of providing their location),complicates this notion. Erickson labels the use of geographic locations in tweets

    associolocative topography because putting a spotlight on the events and activities of

    certain locations, community members construct a shared image of a place through thecontent of their posts (Erickson, 2010, p. 1202). Event commentary in conjunction withspecific geographic locations contributes to the success of tweets in marketing events to

    information seekers.Since UNC-based posts related to events will be the focus of this study, it is

    useful to examine how events broadcast through Twitter are marketed. Jansen et al.(2009) explore the influence of microblogging on brand recognition, trust, and

    attachment, but their conclusions are also applicable to the success of tweeted events.This model of the interaction between microblogging and brands can be altered and

    utilized in our current study, where "brand" is replaced with "event".

    Figure: Model of relationships between Twitter posts about events and the successes ofindividual events. (Jansen, Zhang, Sobel, & Chowdury, 2009)

    Using a lecture hosted by the School of Information and Library Science at UNC

    as an example of an event, we can see the perceived level of success for events posted onTwitter to the UNC online community. In the first column "Event Knowledge," planners

    of the event (information providers) will post the event information and increase theawareness and improve the image of the event (how the event is initially perceived before

    Twitter posts). Once the lecture event is tweeted and seen by followers (informationseekers), the "Event Relationship" is altered and the trustworthiness of the information

    source becomes important in determining whether followers will percieve value based onpast events or past experiences with the information provider. The physical event is also

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    8/15

    included in this column; based on how well the event is attended and/or perceived, usersmay form attachments to the event or those who planned or tweeted it. The "Behavioral

    Outcome" column specifies the success of the event based on the level of attendance andpossibility of attendance at future events. Success of the event can also be mearsured

    based on the amount of tweets and re-tweets following an event, and the perceived

    attitude of those posting the tweets.Other researchers have developed methodologies and algorithms for determininginfluence and authority on Twitter. Guy Hagen's twInfluence is an online tool and API

    that utilizes social network graph theory to examine the "horizon of influence" aroundspecific Twitterers. It is used to measure the "combined influence of Twitterers and their

    followers" (Hagen, n.d.). TunkRank is a similar tool developed by Daniel Tunkelang thatfocuses on a Twitterer's audience. Tunkelang has described TunkRank as "a Twitter

    analog to [Google's] PageRank" (Tunkelang, 2009a). TunkRank assigns a score to eachTwitter user that "is a reflection of how much attention your followers can both directly

    give you and propagate up to you" (Tunkelang, 2009b). Most recently, researchers in theWeb Ecology Project attempted to analyze Twitter influence as a measure of the number

    of replies and retweets a Twitterer is able to generate (Leavitt, Burchard, Fisher, andGilbert, 2009). It is this approach to influence that will inform our own methodology and

    research.

    2.7 Collaborative Storytelling

    Up to this point, we have defined microblogs (specifically Twitter), their users,and their motivations and intentions, but we have not yet discussed the collaborative

    storytelling aspect of Twitter. Collaborative storytelling through social networking siteshas been explored briefly by others, but without a specific focus on the event-based

    stories produced and promoted through Twitter.An examination of Alternative Reality Games (ARGs) and the use of online

    collaborative storytelling tools is the focus of one study with similar implications to ours.Kim, Lee, Thomas, and Dombrowski (2009) performed case studies on five popular

    ARGs to examine how online communities were formed and maintained, and definedhow these communities solved problems collectively as a new genre of collective

    storytelling. The driving force behind ARGs "is to bring groups of people together,online and physically, in sharing and shaping a story" with the help of new collaborative

    storytelling devices such as blogs, wikis, and microblogs (Kim et al., 2009, Introduction,para. 9). One of the case studies included in the article was Year Zero (2007), which was

    started as a promotional tool for a new album by the band Nine Inch Nails. Participantshad to collaborate both online and through physical meetings in order to solve the

    problem presented by the game, a process through which real communities andrelationships were formed (Kim, Lee, Thomas, & Dombrowski, 2009). ARGs are, of

    course, a single example of how online storytelling connects physical venues and virtualcommunities through the use of Web 2.0 technology

    With similar motivations to the Kim et al. study, our proposed study seeks toexamine how stories about events at UNC are told through Twitter. This study will

    increase understanding of how communities of individuals can use tools like Twitter toform connections, learn new information, and further participate in the online storytelling

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    9/15

    milieu. Much of the current literature regards visual storytelling (e.g. through the photo-sharing SNS Flickr.com). Ultimately, we believe that this study will contribute to the

    burgeoning field of online text-based storytelling.

    III. Methodology

    3.1 Introduction

    To understand how public, collaborative stories about events are constructed on

    social networking sites, we propose a study that collects and analyzes data fromTwitter.com about events at UNC Chapel Hill. Twitter is an ideal platform for our study

    because of the large amount of public data available on the site, and because of theexistence of tools that can collect and help analyze that data.

    Using the data collected from Twitter, we endeavor to explore the nature ofauthority in online collaborative storytelling. With the recent addition of the entire

    Twitter backlog to the Library of Congress, seemingly ephemeral tweets will become a

    source of valuable context for historians (Raymond, 2010; Stross, 2010). While it isunknown exactly how researchers of the future will utilize Twitter, it is at least plausiblethat tweets could become part of the context that makes up the stories of public events. It

    is important, therefore, to understand the nature of authority and influence incollaborative storytelling environments like Twitter; the voices that have the most

    influence in the stories of our lives may very well be the voices that shape history.

    3.2 Research Goals

    The efforts to define authority on Twitter have so far centered around the notionof global influence. However, we believe that stories about scheduled events, told by the

    people who were there to experience them, are worthy of separate consideration becauseof their historical implications. We hypothesize that events related to information

    workers (e.g., faculty, staff, and students associated with campus libraries and the Schoolof Information and Library Science) are more likely to be tweeted about than other

    events. The net effect of such a phenomenon would be that events on Twitter related toinformation workers are over-represented when compared to other events.

    We also hypothesize that certain perspectives are privileged over others onTwitter. Specifically, we hypothesize that tweets about events by information workersare more likely to be "retweeted" than tweets about events by other UNC community

    members. Because a retweet occurs when users re-post the tweets of others, a post that isretweeted can be said to reflect a privileged perspective, i.e., a perspective that is valued

    more by Twitter users (Leavitt, 2009). We believe that tweets by information workersare more likely to be retweeted because their networks are more likely to value Twitter as

    an information sharing platform. If our hypothesis is true, it could mean that storiesabout events on Twitter are biased towards the perspectives of information workers.

    Our research questions are:

    1. Who are the users telling stories about events at UNC on Twitter? We are interested in

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    10/15

    how often the posts of information workers are retweeted compared to the averageTwitter population. Are the posts of information workers retweeted more often, i.e., are

    their stories over-represented on Twitter compared to other types of campus events? Whois telling the stories on these websites?

    2. What kind of events have their stories told on Twitter? We are interested indetermining which events get tweeted about. How do popular stories on Twitter compareto other ways of telling stories, i.e., are events on Twitter the same events that get

    covered by the campus newspaper of record, The Daily Tar Heel?

    3.3 Steps Followed and How Results Would Be Analyzed

    We will use the online tool ContextMiner for data collection. Twitter'sapplication programming interface (API) allows ContextMiner to collect sophisticated

    data about users' posts to Twitter. According to its developer Chirag Shah, ContextMiner"is a framework to collect, analyze, and present the contextual information along with the

    data...[it] provides tools to collect data, metadata, and contextual information off the Webby automated crawls" (Shah, n.d.). Currently, ContextMiner does not fully support the

    functionality required by our proposed study. However, the source code forContextMiner is offered under a Creative Commons license, and can be adapted for the

    study outlined below. Preferably, the researchers will work directly with Shah to extendthe core functionality of his software.

    The primary advantage of ContextMiner is that it allows for the creation of"campaigns" around specific search queries. For example, a campaign could be created

    that collects tweets returned from the query "UNC graduation." This mechanism willallow us to monitor the stories being told about specific UNC events. The campaign can

    be set to collect new tweets every day, which would allow us to see how stories unfoldover time. Several data points are collected for each tweet in the search results. The

    same data points will be collected for replies to and retweets (i.e., repostings) of theoriginal tweets:

    - the plain text itself- the Twitter URL of the tweet

    - the author- any outside links referenced

    - the timestamp

    The effective formulation of search queries is essential to gathering meaningfuldata with ContextMiner. This study's search queries are selected using information found

    on the official UNC-CH online calendar at events.unc.edu. This calendar is the mostcomplete representation of UNC-CH related events known to us. Two factors determine

    how we must form our search queries using the data from this calendar: the varied waysin which events are listed, and the requirement that Twitter posts be less than 140

    characters.In the "month" view of the calendar, events are listed in various ways, including:

    event titles; event descriptions; the names of focal persons and/or groups; titles oflectures, workshops, and seminars; or some combination thereof. Even events of the

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    11/15

    same type (e.g., all film screenings) are not always listed in the same way. Because ofthis diversity, the event listings do not suggest an obvious way to formulate

    ContextMiner queries in such a way that we are likely to get results that are related tothose events. For example, if we always used the event title as the search query, the

    resulting data set would be incomplete. Tweets about a lecture may be more likely to

    include the name of the speaker rather than the title of their talk, so searching by lecturetitle would be insufficient.A unique aspect of Twitter is that posts to the platform must be comprised of 140

    characters or less. This limits the way in which users can describe events on Twitter. Forinstance, the event title "Reflections on the Lumbee Battle for Federal Recognition: The

    Future of Self-Determination for Native American Tribes - Arlinda Locklear, Esq." isitself longer than 140 characters, so Twitter users would be unable to describe the event

    on Twitter using the full title. In general, a Twitterer uses the least amount of characterspossible to identify that a post is about a specific event.

    These two factors require that the researchers phrase their ContextMiner searchqueries in very broad and extremely concise terms to ensure that tweets about a campus

    event are harvested. Because such queries would also produce an inordinate amount ofdata, the searches in this study are further limited to tweets that originate within Chapel

    Hill. This will help ensure that a large number of the results are tweets by people actuallyattending an event on the UNC-CH campus, and not about unrelated events elsewhere.

    With these factors in mind, the full text of the webpage events.unc.edu isharvested. After removing numbers and stop words (the, PM, cont, etc.), every unique

    word or string becomes a search query that will be used to start a campaign inContextMiner. Unfortunately, even after limiting our data collection to tweets that

    originate in Chapel Hill, we will undoubtedly collect tweets that are not relevant to UNCevents. The resulting data set will be hand coded by at least two members of the research

    team in order to ensure relevance of the data and cross-coder reliability.Our analysis of data will focus on the establishment of ratios that will illustrate

    the nature of authority in storytelling on Twitter. All comparisons must be made betweenevents with a similar capacity for attendance - a football game is more likely to be

    tweeted about than a guest lecture because more people are likely to attend a footballgame. We will use the classroom seating capacity data at http://hotline.unc.edu in order

    to compare events of similar sizes.

    3.3.1 Significance

    In order to test our first hypothesis, we must establish a baseline measure of what UNCevents are significant to the Twitter network. This is accomplished by comparing the

    number of events that get tweeted about to the total number of UNC events:

    significance of UNC events to Twitter = # of events that get tweeted / total # of UNCevents

    This percentage gives us a proper context to actually analyze our data to test our first

    hypothesis. This is accomplished by comparing the events related to information workersthat are tweeted to the total number of events tweeted. An event is related to information

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    12/15

    workers if it is sponsored by any of the campus libraries or the School of Information andLibrary Science (SILS):

    significance of information worker related events to Twitter = # of tweeted information

    worker related events / total number of tweeted UNC events

    3.3.2 Influence

    Our analysis of influential Twitterers will use the measure of influence proposed by theWeb Ecology Project (Leavitt, 2009). This measure focuses on the actions generated by

    a post: its replies and retweets. A Twitterer is an information worker if their real nameresolves to some connection to SILS or campus libraries in the public UNC LDAP

    directory.

    i1 = # of retweeted posts by information workers at UNC / total # of retweeted posts byTwitterers at UNC

    i2 = # of replied-to posts by information workers at UNC / total # of replied-to posts by

    Twitterers at UNC

    information workers influence = i1 + i2 / 2

    3.4 Benefits and Limitations of Methodology

    As we have seen in the literature review, there are many unanswered and, moresignificantly, unasked questions around the practice of online storytelling. However, any

    study that aims to collect detailed quantitative data about Twitter usage will be limited bythe parameters of the social network site's application programming interface (API). In

    the case of this proposed methodology, it is the program ContextMiner that is bound bythese limitations. Therefore, the questions that we might be able to answer convincingly

    with the data we expect to collect are rather specific.Additionally, the fact that Twitter posts are limited to 140 characters makes the

    formulation of search queries very difficult. Twitterers often develop creative languagein order to overcome this limitation. It is possible that our search queries will miss some

    posts about events because of this.Finally, our measure of influence depends on being able to match UNC Twitterers

    to their professions by using the real name field of their Twitter profile. It is possiblethat some Twitter users discussing events on campus do not use their real names for

    privacy reasons, or that their real names will not successfully map to the LDAP directory.

    3.5 Future Study

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    13/15

    A similar study to ours should be conducted to gather collaborative storytellingdata about other social networking sites such as Flickr and YouTube. Currently,

    ContextMiner capabilities are concentrated on collecting data about videos on YouTube.In fact, the data that can be collected from YouTube is much richer than the data one

    could collect with ContextMiner from Flickr and Twitter. Because the resultant data is

    more complex, more questions could be asked of it. However, asking the same questionsof Flickr and YouTube that we ask of Twitter would allow researchers to compare andcontrast the nature of storytelling across the various SNSs.

    By having data about storytelling SNSs that primarily employ one medium (text,image, or video), one might be able to move towards a theory of media affordance in

    online storytelling. What kinds of stories tend to be told through text? Image? Video?What populations use text more than either image or video to tell their stories? If certain

    populations are telling fewer stories with video, does that mean that they have chosen notto, or does it mean that the more expensive medium of video is still out of reach as a

    storytelling tool for low-wealth populations? There are many aspects with which futureresearch could expand on this current study.

    IV. Conclusion

    The use of online social networking applications has grown steadily over the pastdecade, and by all indications will continue to become an increasingly larger influence on

    social interaction and collaborative storytelling. Microblogs, such as Twitter orFacebook status messages, will provide future researchers with an almost unprecedented

    view into our daily lives. As more and more communication and information seekingoccurs in this medium, it will become increasingly important to study the text and

    metadata associated with microblog posts. This study will lay the ground work for manyother similar studies in the future.

    Throughout our research we have found articles related to issues such as Twitter,blogging and microblogging, informal communication, electronic word-of-mouth, and

    collaborative storytelling. This proposed study would bring together these separate issuesto create a more overarching view of how collaborative storytelling evolves before,

    during, and after an event of interest, as well as who is considered an authority oninformation related to the event. The information gain from this study would be

    invaluable to information workers. By better understanding how information isdisseminated and transmitted in the virtual world, information workers will be able tomaximize their influence and the effectiveness of their information dissemination.

    By studying the metadata associated with tweets rather than just the 140characters that were tweeted, as well as the web of communication created through

    retweets and replies, we will be able to gain a much deeper understanding of an event.We will be able to know not just that an event occurred, but who attended, how they

    found out about the event, why they attended, what they thought of the event, and somany more nuggets of information that expand the story of an event. Twitter asks the

    simple question What are you doing? By studying the aggregate of tweets related to anevent, we hope to answer the broader question What are we doing?

    V. Works Cited

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    14/15

    Cheung, C. M., & Lee, M. (2010). A Theoretical Model of Intentional Social Action inOnline Social Networks.Decision Support Systems, 49 (1), 24-30.

    Ennew, C. T., Banerjee, A. K., & Li, D. (2000). Managing word-of-mouth

    communication: Empirical evidence from India.International Journal of Bank

    Marketing, 18 (2), 75-83.

    Erickson, I. (2010). Geography and Community: New Forms of Interaction Among

    People and Places.American Behavioral Scientist, 53 (8), 1194-1207.

    Granovetter, M. D. (1973). The strength of weak ties.American Journal of Sociology,

    78 (6), 1360-1380.

    Hagen, G. (n.d.). Twitter Influence Analyzer. Twinfluence. Retrieved April 18, 2010,from http://twinfluence.com/about.php

    Hastings, R. (2009). Collaboration Tools, 2.0 Style.Library Technology Reports, 45 (4),19-27.

    Hennig-Thurau, T., Gwinner, K. P., Walsh, G., & Gremle, D. D. (2004). Electronic word-

    of-mouth via consumer-opinion platforms: What motivates consumers to articulatethemselves on the internet?Journal of Interactive Marketing, 18 (1), 38-52.

    Kim, J., Lee, E., Thomas, T., & Dombrowski, C. (2009). Storytelling in new media: The

    case of alternative reality games, 2001-2009.First Monday 14 (6).

    Jansen, B. J., Zhang, M., Sobel, K., & Chowdury, A. (2009). Twitter Power: Tweets as

    Electronic Word of Mouth.Journal of the American Society for Information Science andTechnology, 60 (11), 2169-2188.

    Java, A., Song, X., Finin, T., & Tseng, B. (2007). Why We Twitter: Understanding

    Microblogging Usage and Communities.Joint 9th WEBKDD and 1st SNA-KDDWorkshop '07. San Jose: Association for Computing Machinery.

    Leavitt, A. (2009, March 19). Serendipity, or Twitter and the Narrative of Rhetoric.

    Department of Alchemy. Retrieved April 15, 2010, fromhttp://doalchemy.org/2009/03/serendipity-or-twitter-and-the-narrative-of-rhetoric/

    Leavitt, A., Burchard, E., Fisher, D., & Gilbert, S. (2009). The Influentials: NewApproaches for Analyzing Influence on Twitter. Web Ecology Project. Retrieved April18, 2010, from http://www.webecologyproject.org/2009/09/analyzing-influence-on-

    twitter/

    Levin, D. Z., Cross, R. L., & Abrams, L. C. (2002). The strength of weak ties you cantrust: The mediating role of trust in effective knowledge transfer.Best Papers

    Proceedings of The Academy of Management.

  • 8/4/2019 What Are We Doing? A Study of Collaborative Storytelling Through Twitter

    15/15

    Nardi, B. A. (2005). Beyond bandwidth: Dimensions of connection in interpersonalcommunication. Computer Supported Cooperative Work: The Journal of Collaborative

    Computing, 14 (2), 91-130.

    Nardi, B. A., Schiano, D. J., Gumbrecht, M., & Swartz, L. (2004). Why we

    blog. Communications of the ACM, 47(12), 41-46.

    Palen, L. & Dourish, P. (2009). Unpacking "Privacy" for a Networked World. CHI

    Letters, 5 (1), 129-136.

    Raymond, M. (2010, April 14). How Tweet It Is!: Library Acquires Entire Twitter

    Archive. Library of Congress Blog. Retrieved April 19, 2010, fromhttp://blogs.loc.gov/loc/2010/04/how-tweet-it-is-library-acquires-entire-twitter-archive/

    Reichelt, L. (March 2007). http://www.disambiguity.com/ambient-intimacy/

    Shah, C. (n.d.). ContextMiner - About. ContextMiner. Retrieved April 30, 2010, fromhttp://www.contextminer.org/about.php

    Stross, R. (2010, April 30). When History Is Compiled 140 Characters at a Time. TheNew York Times. Retrieved from

    http://www.nytimes.com/2010/05/02/business/02digi.html?hpw

    Tunkelang, D. (2009a, January 13). A Twitter Analog to PageRank. The Noisy Channel.Retrieved April 18, 2010, from http://thenoisychannel.com/2009/01/13/a-twitter-analog-

    to-pagerank/

    Tunkelang, D. (2009b). TunkRank :: About. TunkRank. Retrieved April 18, 2010, fromhttp://tunkrank.com/about

    Twitter, About. (April 30, 2010). Retrieved from http://twitter.com/about

    Zhao, D., & Rosson, M. B. (2009). How and Why People Twitter: The Role that Micro-blogging Plays in Informal Communication at Work.Proceedings of the ACM 2009

    international conference on Supporting group work(pp. 243-252). Sanibel Island:Association for Computing Machinery.