Self Organizing Content

Team members

Matteo Cernison, Simeona Petkova & Thomas Poell

Introduction

Social media have their own specific organizing principles, which are implemented by the users themselves. Facebook has its “like button”, Flickr and Del.icio.us work with tags, while Twitter is organized through hashtags and retweets. The question is: how well do these principles organize content? Is there something like ‘wisdom of crowds’?

Clay Shirky certainly thinks so. His slogan is: “The Web has an editor, it's everybody.” He maintains: “by letting users tag URLs and then aggregating those tags, we're going to be able to build alternate organizational systems, systems that, like the Web itself, do a better job of letting individuals create value for one another, often without realizing it (Shirky 2005, 18) .” The idea is that through collaborative tagging, involving large numbers of people, very precise descriptions for particular objects will develop. In fact, it would allow for the development of alternate descriptions for a single object, depending on the group which is tagging.

By contrast, Jan Simons is much more careful. He has argued that free tagging systems are prone to be highly imprecise because of various semantic problems (same word meaning different things), different levels of categorization, and spelling mistakes. In turn, this makes the emergence of effective crowd sourced organizational systems rather unlikely (Simons 2008).

So far, this debate has mainly focussed on the description of individual web objects. According to Shirky this is also the main advantage of tagging over traditional classification systems, as tagging allows for the classification of individual objects, instead of aggregating sets of objects into broad categories. In this research, we are precisely interested in the aggregations of objects. More specifically, we want to find out whether actual accounts of an event or issue emerge through collaborative organization practices.

To address this question we specifically examined how the Ground Zero Mosque issue, which is not actually a mosque and not on Ground Zero, but an Islamic center in lower Manhattan, is organized on Twitter. Do the main organizing principles on Twitter, hash tags and retweets, effectively aggregate the tweets on the Ground Zero Mosque issue?

We focussed on the ‘Ground Zero Mosque’ issue, as it is fresh and appears open to conflicting interpretations. At least, this is what the NYtimes and the Foxnews headlines suggest.

Research questions

We have operationalized this question through the following research questions:

• Where can we find GZM issue on Twitter? What are the Twitter queries to access the issue? What are the dominant queries?
• How many of the tweets are organized through hashtags and retweets?
• Does a hashtag and its related hashtags allow us to understand what a collection of tweets is about?
• Do the top retweets per twitter query produce a distinct account of a particular collection of tweets?

Method

Where can we find GZM issue on Twitter? What are the Twitter queries to access the issue? What are the dominant queries?

1) We queried the news for the "Ground Zero Mosque" issue, and we determined the various labels circulating in the news on this issue. It turns out that "Ground Zero Mosque" is indeed the most used in the news. Other labels are “New York Islamic Center”, “Muslim Community Center”, which are the labels which were, for example, used by the NYtimes.

2) We queried Twitter for these labels through the Twitter Scraper.

3) On the basis of these results, we determined the specific hash tags for the GZM issue: #GZM, #mosque, #Groundzeromosque.

4) We determined the tweet activity per twitter query.

How many of the tweets are organized through hashtags and retweets?

1) Through the Twitter Scraper Analysis Tool, we determed the hashtags used, and retweets for the six query results.

2) We calculated the total number of hashtags used per Twitter query and divided it by the total number of tweets for that query. This allowed us to calculate the average hash tag activity per tweet for each query.

3) We calculated the total number of retweets per Twitter query and determined the % of retweets for the total number of tweets per query.

Does a hash tag and its top related hastags allow us to understand what a collection of tweets is about?

1) For each Twitter query we determined the top hashtags, and visualized the relative number of times a hashtag was used in each query.

2) We examined what the top hashtags per query were abbreviations for.

3) We compared the top hashtags across queries.

Findings

Where can we find GZM issue on Twitter? What are the Twitter queries to access the issue? What are the dominant queries?

Surprisingly, the general query ‘Ground Zero Mosque’ produced much more results than the hashtag queries. This is a first indication that a major part of the activity on Twitter for this issue is not organized through hashtags. When we examine the related hash tags for each Twitter query, under the second research question, we will look further into this issue.

How many of the tweets are organized through hashtags and retweets?

Most strikingly, we found that there is a big difference between twitter queries in terms of hashtag organization. For the text queries, there was relatively little hashtag organization. Many of the tweets for the textual queries did not contain any hashtag at all: an average of respectively 0,62, 0,42, and 0,26 per tweet (which is particularly low, if we consider that some tweets contained multiple hashtags). By contrast, the hashtag queries contained an average of 3.3 hashtag per tweet. In combination, this suggests that there are at least two types of actors tweeting who organize their tweets very differently.

The difference in retweet organization are much less distinct. What is striking is that the ‘Fox’ type labelling “Ground Zero Mosque” are in terms of % (10 tot 20%) retweeted more than the NYtimes type labelling.The most retweet activity was organised through the #GZM.

Does a hashtag and its top related hash tags allow us to understand what a collection of tweets is about?

The most prominent finding is that the most queries do not produce radically different related taggings of tweets generated by these queries. For all queries, except one, the top related hashtag is #TCOT (Top Conservatives on Twitter). This suggests that the GZM is very much a conservative issue (it does not mean that the “Top Convervatives” dominate the Twitter space). This is also denoted by the presence of the hash tags: #teaparty, #gop, and #ocra (organized conservative resistance). Obama is clearly much less of a presence in the GZM issue on Twitter. The only query which is hashtaged radically differently is the “New York Islamic Center’ query. This query is strongly dominated by the news, especially CNN.

Overall, the differences are not especially strong between queries, which suggests that is difficult to distinguish between different accounts of an issue on the basis of related hashtags.

<!--EndFragment-->

Do the top retweets per twitter query produce a distinct account of a particular collection of tweets?

The top retweets much more clearly than the related hashtags appear to organize distinctly different accounts of a particular issue. This becomes immediately clear, when we examine the keywords appearing in the top retweets per query.

The stories based on the keywords from the RTs:

1. “Ground Zero Mosque” : Coke Ground Zero, tolerance, intolerance

1. #GZM : Muslim Terrorists, Obama, Israel, US government funds

1. #mosque : stabbed, muslim, sad, media frenzy, implications

1. #groundzeromosque : NY issue, Pearl Harbour, Fox News, exposing, terrorist, White House, Obama

1. “New York Islamic Center” : Protesters, rally, against, controversial

1. “Muslim Community Center” : Muslim community; happier, debate, anti-muslim, honoring, monuments

Thus, this project suggests that especially the top retweets allow us to distinguishing between particular accounts of an issue in the Twitter space. The top related hashtags are much less discriminatory in this respect.

Implications for Further Research

One way to extend the project further would be to consider the new implementations in the way Twitter organizes content such as Related Tweets. Users will be able to see tweets with the same hashtag, or in the same conversation, or in the same place. However, there is still the question whether the related tweets will organize content (more) effectively.

References

Shirky, Clay. ‘Ontology is Overrated: Categories, Links, and Tags’. Clay Shirky’s Writings About the Internet, 2005. <http://www.shirky.com/writings/ontology_overrated.html> (accessed August 15, 2010).

Simons, Jan. ‘Tag-elese or The Language of Tags’. Fibre Culture Journal, issue 12, 2008.