Welcome to the 12th course lecture for COM/SOC 375: the Social Networks class of the University of Maine at Augusta. This week continues last week’s discussion of online networks as we listen in on a webinar by NodeXL founder Marc Smith; watch for his visualization tips and description of the six major kinds of online networks. We also will begin to think about ways to describe the content, not just the structure, of social networks. In your homework this week, I’m asking you to veer away from a pre-packaged assignment and mine Twitter for your own hashtag gold in your own way. Finally, for this week’s participation, I’d like you to predict the future of social media; will the past predict the future, or are we about to veer off onto a new course?
Before you start this week’s lecture, be sure to read Chapter 10 in your Hansen textbook. Also be prepared to set aside some time for listening to audio and watching a webinar; Rory Cellan-Jones’ podcasts may be just the thing to listen to during the week’s commutes. In this lecture, we’ll consider the following subjects in a combination of text, audio, images and video (be sure to review them all):
- Rory Cellan-Jones: The Past (and Future?) of Online Social Networks
- The Founder Speaks: Marc Smith on NodeXL and the Six Types of Online Network
- Analyzing Content: Top Tens, Semantic Networks and a Rehash of Hashtag Networks
- Finding Networks in Shakespeare
- Assignment: Chart Your Own Twitter Search Network
Rory Cellan-Jones: The Past (and Future?) of Online Social Networks
The online websites devoted to interaction that we now call “social networks” didn’t simply spring out of a genius’ imagination one day; they developed over a number of years through in a process of creative collaboration. Follow this link to BBC Radio 4 and listen to all three episodes (30 minutes each) of Rory Cellan-Jones’ 2011 radio program “The Secret History of Social Networking.”
After you’ve finished listening, consider what you’ve heard in the context of what you know now. How much has changed in three years? The WELL, AOL, MySpace and bulletin boards mark the charming past of online social networks. Facebook, Twitter and Pinterest mark the current mainstream. What does the future hold? What comes next? Using this Padlet (or the lecture’s blog comments section), forecast the future in a thoughtful post that is based on careful observation of trends today. (And pssst… be original. Be sure not to repeat what someone else has already foretold. This means that those who jump in late may find the task more difficult.)
The Founder Speaks: Marc Smith on NodeXL and the Six Types of Online Network
Sociologist Marc Smith is the Director of the Social Media Research Foundation and one of the authors of the NodeXL visualization software that we have been using for social network analysis and visualization during the semester. In a “webinar” produced last fall, Smith describes his motivation behind the development of NodeXL software, discusses the relevance of NodeXL network analysis for business and non-profit enterprises, and finally shows off some nifty visualization tricks that can produce stunning images of online social networks. Feel free to skip the first ten minutes to jump in to the meat of the presentation:
When you track a hashtag this week, can you tell me which of the six types of network you’ve uncovered?
Analyzing Content: Top Tens, Semantic Networks and a Rehash of Hashtag Networks
People don’t flock to social media in order to appreciate its social network structure. People share meaning over social media, and for that reason the content of social media messages matter. Content can be analyzed, too, and the basic tools of NodeXL can be combined in a number of ways to accomplish this. The video embedded below, running about 45 minutes in length, walks you through three ways to uncover meaningful patterns in the content of Twitter posts through NodeXL software:
- Method 1: the built-in “Top Ten” metrics function of NodeXL that quickly and easily summarizes the most salient web pages, keywords and semantic connections in any Twitter discussion.
- Method 2: using the “word pairs” option in NodeXL metrics to generate a full-fledged semantic network, in which words are now the nodes and closeness in a Tweet is the link.
- Method 3: Hashtag networks. We’ve already covered the general territory of hashtags in our previous week’s lecture, but can hashtags be nodes in their own networks? Absolutely. Hashtags are claims about the important subject or bent of a social media post. When two hashtags appear together in a Twitter post, there’s been an implicit claim that the subjects of those hashtags are linked.
Finding Networks in Shakespeare
In the lecture section above, we looked at the “word pairs” tool in NodeXL as a way of drawing out connections between concepts in the content of social media posts. We drew content from a “spigot” — an automated process using the “Import” command in that program. What’s great about NodeXL’s “word pairs” tool, however, is that it can take content from just about anywhere, from just about any time. How far out can we go? Why not Elizabethan England, 414 years ago? That’s where and when the comedy Much Ado About Nothing was written by William Shakespeare. Consider this exchange from Act I:
Beatrice. I wonder that you will still be talking, Signior Benedick: nobody marks you.
Benedick. What, my dear Lady Disdain! are you yet living?
Beatrice. Is it possible disdain should die while she hath such meet food to feed it as Signior Benedick? Courtesy itself must convert to disdain, if you come in her presence.
Benedick. Then is courtesy a turncoat. But it is certain I am loved of all ladies, only you excepted: and I would I could find in my heart that I had not a hard heart; for, truly, I love none.
Beatrice. A dear happiness to women: they would else have been troubled with a pernicious suitor. I thank God and my cold blood, I am of your humour for that: I had rather hear my dog bark at a crow than a man swear he loves me.
Benedick. God keep your ladyship still in that mind! so some gentleman or other shall ‘scape a predestinate scratched face.
Beatrice. Scratching could not make it worse, an ’twere such a face as yours were.
Benedick. Well, you are a rare parrot-teacher.
Beatrice. A bird of my tongue is better than a beast of yours.
Benedick. I would my horse had the speed of your tongue, and so good a continuer. But keep your way, i’ God’s name; I have done.
Beatrice. You always end with a jade’s trick: I know you of old.
The in-person exchange between characters back then might be an online exchange in today’s world. How can we visualize the connections between the concepts in it?
Step 1: Enter data into NodeXL manually. In this case, I have chosen to use three separate NodeXL files. In the first one of these files, I have entered all communications by Beatrice to Benedick in the “Edges” tab, using an “Other Column” that I have named “text”:
Thanks to computers, entering this text is relatively easy. I’ve simply cut (Control-C) and pasted (Control-V) each line in to a separate row, indicating a separate piece of communication. For each row above, Beatrice is in the “Vertex 1” column as the “sender” and Benedick is in the “Vertex 2” column as the “recipient.” In the second NodeXL file, I have similarly entered all communications by Benedick to Beatrice:
Step 2: Generate word pairs. In each of these NodeXL files I have used the metrics commands to generate lists of word pairs for the . Because relatively few words are spoken between the pair in this conversation, I have made sure to uncheck the box “Skip words and word pairs that only occur once.” I’d like every verbal connection to be apparent.
The results look like this, with one set of word pairs for Benedick’s communications to Beatrice and another set of word pairs for Beatrice’s communications to Benedick:
Once you see these word pairs, you may want to do some manual editing. For example, you’ll notice in the semantic network I visualize below that the Shakespearean word i’ is absent; this is a connecting word short for “in,” and these connecting words are typically left out of a semantic network. Also note that the various forms of the word “love” (love, loves, loved) have been combined into one semantic concept, “love.”
Step 3: Copy word pairs to new NodeXL worksheet. In this third NodeXL file, word pairs from Beatrice to Benedick and from Benedick to Beatrice are placed in the “Edges” tab, with the pairs of words placed in “Vertex 1” and “Vertex 2” column. A column named “speaker” is added in the “Edges” tab. The “speaker” column is labeled as either Beatrice or Benedick, depending on who is doing the speaking; in order to distinguish between these in our network graph, the “color” column labels the Beatrice’s word pairs as red edges in the graph and Benedick’s word pairs as blue edges:
The overall result looks like this as a static picture:
But there’s one more step we could take. The lines in a Shakespeare play are not simply revealed to the reader all at once. Rather, they are read in sequence, and part of the joy of a Shakespeare comedy is to watch meanings of words and phrases being volleyed back and forth, twisted, and returned to in new ways. In other words, the semantic network of meanings in a Shakespeare play is dynamic (changing over time), not static (remaining the same over time).
To capture the dynamic nature of this network, I add a new column called “line” to the “Edges” tab of the third NodeXL file. Each row has a value of “line” from 0 to 10, with the lowest number (0) indicating the line in the play that is spoken first, the highest number indicating the line that is spoken last (10), and intermediate numbers indicating the order of the lines.
You’ve read about “filtering” in your Hansen textbook; let’s apply that knowledge now. By selecting the “Dynamic Filter” option in the “Document Actions” window of NodeXL, you can choose which edges to show for different values of the column “line”:
Start by only viewing the edges in the semantic network for the first line, and use the sliding bar in the “Dynamic Filters” to reveal each line in turn. Try taking a “screen capture” (a video picture of your computer screen) while doing this — a free tool for video screen captures is at screencast-o-matic.com. When you’re done, congratulations: you’ve created a dynamic semantic network of your content:
Of course, the literature of William Shakespeare is not the main object of study in this course. Social networks are. But connections between words can mean just as much as connections between people. And just as in a Shakespeare play, if less eloquently, the characters who use social media trade barbs and banter online, often sequentially. Could you create a similar dynamic semantic network depicting the development of discussion as someone posts to Facebook or a blog and visitors leave comments? Absolutely.
Assignment: Chart Your Own Twitter Search Network
Last week, you engaged in a guided use of NodeXL’s Twitter Search function to generate a Twitter network for the hashtag #mepolitics. This week, I’d like you to conduct and report on your own search, generating your own Twitter network. Click here to learn more about how to search Twitter.
Choose a hashtag, a user, a list or a snippet of text to search through and try your own search, limiting the results to either 500 or 1000 “Tweets.” Take some time to experiment with different searches until you find a Twitter network interesting to you. Use the visualization options in NodeXL to look at the structure of conversation and decide what’s interesting.
Having found an interesting Twitter network, create a meaningful sociogram in which you use space, node design and edge design to depict patterns in who is having Twitter conversations with whom. Include this sociogram in your assignment.
Referring to Chapter 5 of the Hansen text and your work earlier this semester in measuring network characteristics, use the NodeXL to measure as many network characteristics regarding your Twitter network as possible (using your installation of NodeXL Basic or NodeXL Pro). Be sure to report these network characteristics in your assignment.
Referring to the sociogram and network characteristics where appropriate, discuss your choice of a Twitter search and patterns in the resulting network that you obtained. What of importance do these results indicate?
Post a word processing document containing the above information to the appropriate Blackboard Homework section labeled “Homework #8″, and you’ll be all set.