« Multimedia Reporting Examples | Main | News Gathering and Mobile Technologies »
Assignment #2: Information Diffusion in the Blogsphere
By Irfan Essa | January 8, 2008
Information Diffusion in the Blogosphere
Assigned: 1/29/2008 Due: 2/19/2008 before class
Work in Pairs! Get started early, this is a long assignment involving both media analysis, design, and programming.
The web as a publishing platform facilitates the rapid review and re-publishing of information through different outlets such as blogs and podcasts. In particular a characteristic of blogs is that they tend to quote or paraphrase information from traditional news sources and add additional commentary. In fact, 55% of bloggers say that they make a post often or sometimes because of something they read or see in the news (http://www.pewinternet.org/PPF/r/186/report_display.asp). As the information from the original source diffuses through the internet it can be subtly altered, misinterpreted, or outright changed. A microcosm of the process is easily seen in the child’s game of “pass it down”, where each child passes a phrase on to the next in turn. By the time the phrase reaches the last child in the classroom the message has significantly changed.
The purpose of this assignment is to observe and critically analyze how information diffusion occurs with online news sources and then use this understanding to synthesize something new. Using a blog indexing tool such as Technorati (www.technorati.com), BlogPulse (www.blogpulse.com), or Google Blog Search (http://blogsearch.google.com) choose a single news story and track how that story has been referenced and commented upon in the blogsphere (this could involve text blogs, vlogs (video blogs), or podcasts). BlogPulse also has a tool called the conversation tracker (http://www.blogpulse.com/conversation) which can help show the tree of comments originating around an article / topic. You should consider at least the first level of blogs linking to the news source, but may also consider multiple levels of linking. NYTimes provides a list of most blogged articles on its homepage so this is one potential starting point for finding an actively blogged about article.
Write a 2 page analysis (~1000 words) considering the following questions as well as any other issues you find interesting and relevant. Give a short summary of the story. What are your general observations about the interaction between news and the blogosphere? What are the different perspectives that people bring to the story? Can they be classified, and if so, how? Has re-interpretation of the story by bloggers affected the tone and understanding of those commenting on the story? How has the story changed (has it?)? What kinds of tags are used to describe the story and/or posts? Are they appropriate? Finally, how might you aggregate, organize, structure, and re-present the complete story (original story, blogs, blog comments, tags, etc.) for this specific story so that readers would gain a better understanding of the context and varied perspectives? This is meant as a divergent, creative exercise in exploring the range of designs that you could produce. Include sketches for how you would present all of this information in a unified interface.
Now that you have a sense for how information is quoted from news sources in the blogosphere and some ideas for how you might organize that information around the original news items from your interface mockups, generalize these ideas to produce a functional prototype using the Technorati API (http://technorati.com/developers/), or any suitable alternative that you find (please let us know what API you are using). The prototype doesn’t have to implement everything that you indicate in your design, but should be basically functional; proof-of-concept is most important here. Think about how to organize the different perspectives around the information available through the API (e.g. tags / keywords etc.), or how to integrate that data directly INTO the original story through interface / visualization techniques. Additional language processing capabilities are available through packages such as LingPipe: http://www.alias-i.com/lingpipe/, which allows for things like sentiment analysis. Sentiment analysis can be used to categorize blog posts according to their positive / negative affect with respect to the original story. Also, feel free to use any other other package for doing language processing, NLTK is an open source one written for Python, though not as powerful as LingPipe. (http://nltk.sourceforge.net/index.php/Main_Page).
The following paper is a technical look at information diffusion and it may give you some more ideas for the mechanics and models of information in the blogsphere:
- Gruhl D. et al. Information Diffusion Through Blogspace. World Wide Web Conference (WWW). 2004.
If you’re interested in how a site like BlogPulse works, an earlier version of the concept is described in this paper:
- BlogPulse: Automated Trend Discovery for Weblogs. World Wide Web Conference (WWW). 2004.
To Hand In (on T-square)
- Your 2 page write up (include a link to the original story in the write up)
- Sketches for your information design
- A link to your prototype and a .zip file containing your code
Topics: Announcements, Assignments |
Comments
You must be logged in to post a comment.