The focus of this post is to demonstrate how to use and interpret the network visualization feature in Netlytic using tweets about Oscars as a case study.
As Hollywood’s stars gathered to celebrate last year’s finest films in the annual Oscars ceremony, the world watched. This year’s Oscars, which took place on Sunday March 2nd, recorded the highest viewership for the Academy Awards in the last decade with more than 43 millions tuning in to watch the biggest entertainment show.
With the prominence of social platforms particularly Twitter as a tool for immediate conversation around live broadcasts, it’s not surprising that Twitter was the star of the night. According to The Washington Post article, ABC reported that Oscars-related posts on Twitter were up 75 percent from last year, with estimates that nearly 3 million people sent a total of 11.2 million tweets.
In the Social Media Lab, we have been collecting a sample of global tweets using Twitter’s Streaming API which allows us to collect a sample of up to 1% of public tweets. In particular, we were interested in discovering of how much we can learn about the event from just 1% of all tweets? And how nuanced and representative would such sample be?
In total, our sample included ~20M tweets posted between February 28 to March 4, 2014 (a couple of days before and after the event). Out of these tweets, ~111K explicitly mentioned Oscars. For the purpose of this exploratory analysis, we decided to narrow down our focus, and only analyze tweets mentioning Oscars posted on March 2 during the date of the event, (see the spike in Figure 1). In total, our final sample contained 63,430 tweets.
Next we used Netlytic to discover a communication network among Twitter users (see Figure 2). In the network visualization, each dot represents a Twitter user (nodes), lines connecting dots indicate a communicative relationship of who is mentioning, retweeting or replying to whom, the colour represents the sub-community (identified automatically). Netlytic maps social networks using an algorithm, which refers to a specific layout that calculates the placement of each dot relative to other dots. The layout used here is “OpenOrd” and is designed to highlight sub-communities within large networks. It is observed that the network is compromised of a number of distinct groups of clusters.
At the heart of this mishmash of heterogenous colours and nodes, there is one particular cluster that stands out in its centrality and degree of influence. As illustrated in the visualization, the @TheEllenShow cluster (turquoise cluster) is particularly prominent with numerous of nodes connected to it. One obvious answer could be attributed to Ellen DeGeneres, this year’s host, tweeting the most retweeted photo of all time. The Oscar’s selfie was retweeted more than three million times, which surpassed Barack Obama’s ‘Four more years’ retweet record. With such buzz surrounding Ellen’s live tweets, it is worthwhile to take a closer look at this cluster. Figure 3 shows the visual representation of Ellen’s cluster. The cluster shows the extent and intensity of messages surrounding Ellen’s twitter account as most of it were ‘in-degree’ tweets; which means that other twitter users were including @TheEllenShow in their tweets either as direct mentions or retweets.
Another notable cluster category in the network analysis is the one surrounding @TheAcademy. As demonstrated in Figure 4, thousands of twitter users were interacting directly with this twitter account. Since @TheAcademy is the official account for the Academy of Motion Pictures Arts & Sciences, the account was tweeting live updates on the Oscar winners and event’s happenings as it unfolded. This prompted twitter users to engage back through retweets and mentions. Another possible explanation to this condensed cluster around the Academy could be attributed to the #MyOscarPhoto campaign that ran in conjunction with the Oscars. The campaign was an example of the latest User-Generated Content (UGC) stunts that companies are increasingly promoting to create memorable interactive experiences. The campaign’s premise encouraged viewers to follow @TheAcademy, tweet their photos to be featured next to a Hollywood celebrity then watch the TV broadcast on whether they made the cut or not. This is a perfect example of the phenomenon of second screen experience where the line between online and traditional media is increasingly blurred as more people are multitasking and using both mediums to foster a holistic experience around an event.
Another visible cluster of users is around @eonline (see Figure 5). While this cluster is much smaller that the previous two, it is observed that @eonline is at the heart of those interactions with many users mentioning and retweeting their posts. This is the official twitter account for E! Online, the most definitive guide for entertainment news and celebrities gossip. Therefore, the cluster was focused around the fashion and style of celebrities during the Oscars. In addition, this account interacted directly with other official accounts such as @TheAcademy, @TheEllenShow, @Pharrell and a host of celebrity twitter accounts. It is interesting to examine this cluster as it shows the role of other TV stations in promoting and generating buzz around the Oscars. Accompanied with the nonstop tweeting, the station had live coverage from the red carpet as it sent its reporters to interview Hollywood stars and broadcast relevant content on its channel.
This exploratory network analysis has demonstrated how to use and interpret the network visualization feature in Netlytic using tweets about Oscars as a case study. The Twitter communication networks built with Netlytic can be useful in understanding how communities form around specific topics and the extent to which they buzz around niche interests and conversations. In addition, the network analysis demonstrated that some clusters not only interact within themselves but also with other clusters.
Also the Netlytic network visualizer revealed that even with a relatively small sample of tweets, we were able to find some interesting patterns in data. For example, in the case of tweets about Oscars, although the overwhelming majority of the tweets on this topic (~3 million) were simple retweets of the Ellen’s selfie; the network representation of the data shows that there are other topics discussed around this event such as tweets about the MyOscarPhoto campaign and discussions around the fashion and style of celebrities during the Oscars (see above).
Finally, the analysis also begs the question on whether social media is contributing to increased viewership of television or vice versa? While this brief analysis will not answer the question fully, it showcases the interplay between digital and traditional communication platforms. Perhaps, in the not-so-far future, we will not continue to look at each as separate forms of communication and information dissemination. Rather, they will be viewed as complimentary technologies that are used simultaneously.
*By Anisa Awad with contributions from Anatoliy Gruzd and Philip Mai.