User:Htw3/Digital.projects.2009

Overview

edit
  1. Students will create on individual digital project and one group digital project.
  2. The group digital project is due first, during weeks 5, 6, 7, or 8.
  3. The task for the group digital project will be posted to Marc Smith's blog "Connected Action". This contribution will include some type of preliminary analysis related to this course. Examples of an analysis project suitable for publication on Marc's blog might include:
    1. a network analysis of that compares face book networks of the group members
    2. perform or participate in an experiment and write up the results. (Note: we have access to at least two potential experiments).
    3. a description of the network structure among papers or authors that cite a particular publication read in this course.
    4. a description of an online community with interesting collective action dynamics.
    5. a documentation of roles or status dynamics in some group.
    6. a big picture discussion of social change related to an issue raised in the course.
  4. The task for the individual digital project includes projects that build on or extend any of the group projects, or any other digital project that appropriately offers a new contribution in line with the themes of the course. Exact topic, method and presentation is open to discussion. Due during weeks 9 or 10.

Using NodeXL

edit
  1. There is a very helpful document for working with your data in nodexl: http://casci.umd.edu/images/4/46/NodeXL_tutorial_draft.pdf
  2. Please use this section to give advice and tell how to do things as you go.
  3. One example, is that because node size and width are constrained to be between 1 and 10, and counts can easily get bigger than 10, you will want to re-scale those values. one way to do that is to use a formula like:
    1. =(Z+ln(1+X)) where z is an arbitrary minimum value, say 1 or 2. and X is the cell where there count value is stored.
    2. Since ln(20,000) is equal to about 9.9 the natural log is going to put most of the count values we observe in our settings between 1 and 10

Project Idea 1

edit

I've been working on a paper that looks at reader comments left on a New York Times article about LASIK surgery. We would basically be doing a social network analysis of the comments readers left and probably coding the comments (positive, negative, etc.). Depending on how many people we have the actual coding won't take that long and will likely just be copy and paste from Word into Excel (I already have all the data). I'd like to continue on this project further after class for conference submission (University Research Expo and regional conference).

People working on this project: Jeff Kuznekoff (jkuz05@gmail.com),

Comment (Ted): This is a cool project too. It will require some of the same steps as the firefighting project, but will likely want to develop more of a coding scheme for the content of the messages since there are not user profiles.

  • See data collection steps related to the next project for this one as well.

Project Idea 2

edit

Measure and Visualize discussion network in online forum for firefighters. Measure attributes of individuals from their profile pages. Infer edges from content and timing of posts, perhaps measure something about the content too?

The website that I have been looking at is firehouse.com. It has a forum where firefighters and EMS personnel give their opinions and discuss different things. We perhaps could pick a thread, for example one of them is "Taking pictures on a scene", and we could look at how many members contribute to this thread, and who is talking to whom. We could do something similar to what we did in class and put the data in NodeXL. We could collect data from members, such as when they joined the website, how many posts they posted, average posts per day, how many years of firefighting or ems experience.

So if you are interested in working with me on this project let me know.

People working on this project: Dumitru (ds204204@ohio.edu), Alan (88select@gmail.com)

Comment (Ted): I think this is a cool project idea. I see several steps:

  1. Read a variety of threads looking for a couple that seem interesting, and include about 50 posts, or a section of a larger thread with about that many posts.
  2. Write down explicit rules for how you will measure an edge (how you know someone is replying to someone else).
  3. Make a list of all identities who post to your thread (or section).
  4. Go to their profile pages and collect data posted there that seems relevant.
    1. Use a normal excel sheet, and use the user name for the data in the first column and the measure you seek to include from the profile page for the subsequent columns.
    2. after you have recorded all your nodes, with variables about them, use your edge coding scheme to measure the edges in the conversation. Decide if you want to include any other variables related to each message. record those at the time you are recording the edge structure. you might want to include the timestamp for the post, incase you want to show the temporal evolution of the conversation.
  5. Make sure you data is clean, especially make sure that you are consistent in the spelling of the node names. Avoid Ted ted TED situations.
  6. Import or copy past your data into NodeXL. Make sure you do things like define the network as directed, and explore interesting uses of node size, color, shape in order to express the state of variables that might show us something interesting. You can also explore fixing the nodes in place, using autofill columns, etc.

Project Idea 3

edit

Study networks related to a tag or a person on twitter.

I've actually found a few programs that may be of use here, including xefer.com. I am very sad that tweetwheel no longer exists, as it would make the work easier, but I imagine that there is something similar out there.

Alternately, it might be interesting to code and categorize trending topics over the course of several days. what are people tweeting about? Current events? Movies/music? Celebrities? Technology? Something completely random? I'm not really sure I want to get out of this in terms of meaningful data. It could just be interesting, without answering the "so what?" question.

at this point, we are considering using a program called "Trendrr," which can search and graph topics on twitter, google search, and a number of other social media sites. We are interested in the differences between twitter and "traditional" online outlets such as google - are some topics more popular on one than the other? Who picks up on a topic first? How much information is exchanged? Two topics that we have in mind are Michael Jackson's death and the earthquake in Indonesia (alternate topics suggested: H1N1, the typhoon in the Phillipines).

While many of the tweets concerning international events are in other languages, their hash tags are often in English. Still, we are considering running the key words through Trendrr in several relevant languages.

People working on this project: Michelle Calka (mcalka@gmail.com), Katie Kassner (katie.kassner@gmail.com), Joey Argiro (jargiro710@gmail.com)

Project Idea 4

edit

Use data from the etherpad conversation.

People working on this project:

Project Idea 5

edit

Do something related to the SES and Facebook / MySpace divide, related to conversation on etherpad and on blog.

Blog entries: http://group-processes-social-change.blogspot.com/2009/09/facebook-myspace-article.html

http://group-processes-social-change.blogspot.com/2009/09/digital-inequality.html

Etherpad: http://etherpad.com/ALwC63H98U

Specifically: We'd like to go more deeply into other kinds of online social stratification (not even necessarily about race/class). Perhaps each of us can choose a different niche social network such as LinkedIn, Gaia, perhaps online dating sites (OkCupid?), and whatnot, and go deeper from there--like, how do these relate, how are they different, and come up with some sort of big picture conclusion: Is this going to help or hurt how people relate in the outside world? Any ideas are completely welcome, and we're open to whatever directions you want this to take.

We can use Quantcast (http://www.quantcast.com/) to measure demographic information about each of these sites.

http://www.quantcast.com/okcupid.com

http://www.quantcast.com/facebook.com

http://www.quantcast.com/myspace.com

http://www.quantcast.com/linkedin.com

http://www.quantcast.com/gaiaonline.com

People working on this project: Arianna (ariannailiff@gmail.com) Elyse (em141906@ohio.edu) Rachel Reilly (rachelreilly16@gmail.com) Kyle Raffle (kr336008@ohio.edu)

Project Idea 6

edit

Do something new comparing Facebook networks using Touchgraph, or something using Zipskinny, or other ideas introduced in class.

Touchgraph: http://www.facebook.com/apps/application.php?id=3267890192

Zipskinny: http://www.zipskinny.com/

People working on this project:

Project Idea 7

edit

Mark suggested that someone pick up this idea: http://group-processes-social-change.blogspot.com/2009/09/network-example-and-free-project-idea.html

People working on this project:

Project Idea 8

edit

Work with network data and images from usenet groups. Ted has data from old research projects that you can use.

People working on this project:

Project Idea 9

edit

We are interested in exploring the differences in demographics between MySpace users and Facebook users (based on Etherpad discussion). 1. Do socioeconomic statuses seem to vary between the users of both sites? 2.How about education levels, ages, number of children, etc. vary between groups? 3.How does one's involvement in music change their selection/activity on each networking site? 4.When do teens make the switch from MySpace to Facebook generally, and why does MySpace seem more appealing to younger users? These are all questions we are interested in exploring...we're just not sure exactly how we'd like to go about it.

People working on this project: Christina (cg159606@ohio.edu/christina.green.88@gmail.com), Holly (hn320307@ohio.edu), Alyssa (worldwidewaterslide@gmail.com)

Project Idea 10

edit

This project is methodologically similar to project two, only it analyzes conversational patterns within discussion forums for the site Postsecret. It involves codifying and mapping two or three brief chat feeds on socially controversial topics. Due to the anonymity associated with the site, however, involve gathering information form the user's profile page.

Using previous literature on the emergent structure of conversational patterns, we seek to find discrepancies in conversational patterns between neutral topics (car maintenance, coin collecting, etc) and controversial and potentially uncomfortable topics.

People working on this project: Nina Cesare (nlcesare@gmail.com), Jarrid Wong

Project Idea 11

edit

This project is a simulation of group joining behavior. The goal is to create a flexible simulation that can be optimized to match real world network data. A slide show with some preliminary results can be found below:

http://docs.google.com/present/edit?id=0AZ_84r8sQ-wlZGMyZzZiY3FfMzFkenAzdGJmdg&hl=en

People working on this project: Mark Kokoska