Categories

Recent Posts

Okkam - Semantic Web Resource

By Nick | March 12, 2008

I saw this article about a resource the EU is building to help with entity detection / resolution called Okkam. This could be useful for anyone trying to do text processing as a component of their final project.

Topics: Interesting | No Comments »

GUEST SPEAKER: Rasmus Lerdorf (YAHOO!)

By Irfan Essa | March 6, 2008

Rasmus Lerdorf (YAHOO!) was kind enough to come to class and give a guest lecture on class on

A quick introduction to PHP and how it can be used to collect, analyze and distribute information.

His slides for this talk are at http://talks.php.net/show/ygatech5

He covered many things including pointers to

Topics: Guest, Lecture | No Comments »

Final Projects

By Irfan Essa | March 5, 2008

Computational Journalism Final Projects

Out: March 6, 2008

Deadlines:

Details:

Final projects should be composed of both analytic and constructive components; analyze a particular issue and then construct a prototype in response to that. The goal is to build some form of new media artifact or prototype which considers some of the issues that we have been talking about this semester (e.g. journalism practice, automation of reporting, news gathering, sense-making, contextualization, aggregation, personalization, information quality and bias etc.).

Projects should be in groups of three, though depending on the scope of your project you may also have four people in the group (please talk to Irfan or Nick). Post your group and project idea on the swiki as soon as you get it sorted out. Evaluation of the project will be based on an in-class presentation on April 22, a short write-up of the project, and of course the quality and effectiveness of the prototype that you build.

Here are some high level ideas for projects. This is just a small subset of what could be done, so if you have a more creative idea then go for it!

  1. Extend and build upon the “Information Diffusion in Blogosphere” assignment and implement a general purpose News+Blog interface to the varied responses to news articles in the blogosphere. This would also include incorporating text sentiment analysis algorithms or APIs and UI design. What should a blog dashboard look like for an individual story?
  2. Quote De-contextualization Detector: Using the clustered output of a news aggregator such as Google news, collect all of the quotes from those stories and detect if some of them have been shortened or changes. Then allow people online to rate detected altered quotes depending on whether the meaning of the quote is affected in some significant way.
  3. Image Forensics Browser Plugin. There’s been a lot of recent work in the area of image forensics (trying to detect if an image has been altered). See Hany Farid’s page at Dartmouth for more details: http://www.cs.dartmouth.edu/farid/research/tampering.html . The project is to take one (or more of these image processing algorithms) and integrate it into a browser plugin (e.g. A Greasemonkey plugin for Firefox), so that as someone is surfing a news site, then get feedback on the veracity of the images on the page.
  4. Develop a visualization of news information as an interface to either 1) support research and sense-making by a journalist or 2) facilitate news understanding by a consumer or both. You could pick a particular medium of news information such a broadcast/video, print, radio/podcasts, blogs etc. and focus your visualization around the constraints of that medium. You might consider visualizations over time, space or some hybrid form. Think about what visuals would be useful for understanding and answering questions about the data.
  5. Computational Photo Journalism: Taking a textual news story as input automatically or semi-automatically produce a photo essay of that news story using photos taken from news sources or other online photo sources. As an example see NewsInPictures (http://newsinpictures.deakondesign.com/).
  6. Think about how to re-mediate traditional editorial cartoons from newspapers for an interactive platform and then build an online system for either authoring or interacting with these pictorial editorials.
  7. Automatically hyper-mediate multimedia news information for instance by automatically creating links in a textual news story to the audio (or video) portion of the actual interview.
  8. Build an authoring tool of some form to facilitate an element of journalistic practice (e.g. verification of information, contextualization, relevancy, bias managment etc.). In particular you might consider how to build a tool that would make doing good journalism much easier for a citizen journalist. For instance, a tool to easily cut together video interview material into a compelling interview (or group of interviews).
  9. News games (See http://www.newsgaming.com/ for an idea of what this means). Build a newsgame (or several smaller ones) that allow people to explore a complex issue through simulation and gaming. Think about how you might build a news game authoring tool for editorial journalists. Or would it be possible to take a written story and transmediate it into a game automatically.
  10. Collaborative Technologies: 2.0 is all the rage these days. Think about how to leverage the wisdom of the crowds to facilitate journalistic goals. For example: what about a collaborative bias detection system? Or a collaborative image / video debunker?

Feel free to suggest other ideas for the class to consider below.

Topics: Project | No Comments »

Journalism Intern Needed

By Nick | March 5, 2008

I received this request from Barbara Christopher in ISYE for any student interested in getting some experience in the field.

- Nick

Journalism Intern at the H. Milton Stewart School of Industrial and
Systems Engineering

If you’re looking for an opportunity to work as an intern where you will
actually get to report, and work on Georgia Tech’s campus, this may be
the position for you. The average intern should walk away with more than
twenty real reporting clips.

Key Responsibilities: Assist with writing news releases, articles,
promotional materials, and other media-related information about the
Stewart School of Industrial and Systems Engineering, its academic
programs, research centers, the faculty, students, alumni, activities,
events, initiative, etc

Eligibility: A candidate must be an undergraduate student at Georgia
Tech. Have excellent written and oral communication skills. Must be
dependable and creative.

General Information: Interns are expected to work between 15 to 20 hours
a week during the Spring semester and 20 to 30 hours a week during the
summer.

Start date: April 2008 through July 2008

Hourly salary: $15.00 an hour

Application Procedures: Candidates must submit a cover letter, resume,
and writing samples. All application items must be submitted as a
complete package. Incomplete applications will not be reviewed.

E-Mail or fax applications to:
Barbara Christopher
bchristopher@isye.gatech.edu <mailto:bchristopher@isye.gatech.edu>
(404)894-2301

Topics: Announcements | No Comments »

Automated Detection of Image Manipulation

By gtg872p | March 4, 2008

This blog post, entitled “Protecting Journalistic Integrity Algorithmically,” talks about a program that analyzes images in an attempt to locate cloning and other artifacts produced by image manipulation. http://lemonodor.com/archives/2008/02/protecting_journalistic_integrity_algorithmically.html 

Topics: Interesting | No Comments »

George Polk Award for Talking Points Memo Founder

By gtg872p | March 4, 2008

The founder of Talking Points Memo was awarded the first George Polk Award given to a web-only publication. (The Polk is a major award for journalism, and a major honor for the journalist receiving it.) http://www.nytimes.com/2008/02/25/business/media/25marshall.html?pagewanted=all 

Topics: Interesting | No Comments »

Advanced News Summarization

By Irfan Essa | March 4, 2008

Topics: Lecture | 1 Comment »

Geographic Data Maps Mashup

By Nick | February 19, 2008

Geographic Data Maps Mashup
Assigned: 2/19/2008     Due: 2/29/2008 6:00pm

The purpose of this assignment is to create a map mashup / visualization with a newsworthy dataset of your choosing. You’ll find or create some dataset which you think would have some news value and then put it on a map. To do this you’ll need to find or create the dataset, geocode it (if it’s not already), and then display it on a map in some way which facilitates better understanding of the data. Try to think like a journalist: how can you present the data on a map so that an end consumer can better understand it or make better use of it? Keep in mind, each marker on a map can encode not only it’s location, but also additional facets of data through size, color, intensity, etc. Also, Google Maps is more sophisticated than just being able to put individual markers on the map. For instance, you can also mark regions on the map as seen in this example (you have to click on the map). Of course, you can use your choice of map APIs (e.g. Google, Yahoo etc.) to do this. There are a number of resources which you may find helpful through Programmable Web including APIs to potential datasets.

Here’s a map project that was done for last year’s class: http://swiki.cc.gatech.edu/cj2007/uploads/5/Assignment1-Maps.12.html. In this example, the students actually went out and gathering data for each of the map locations (e.g. pictures and text), however we’re not requiring that you do that this year. It’s perfectly OK to find a dataset online somewhere, or to create a small dataset by doing some research online. Datasets could be hyperlocal, national, or even international. Here’s another example from the Sunlight Labs which shows the geographic distribution of federal earmarks: http://sunlightlabs.com/earmarks/.  If you want to go for extra credit, you can also think about mapping 2 datasets which when seen together on the map might provide some additional insight through the juxtapositions of the data.

Some ideas for data to map:


Once you come up with an idea for a dataset, you’ll likely need to do some geocoding to get longitude / latitude data for each point in the dataset. To do this you can use the Google API, or a geocoding API such as Yahoo, Geocoder.us, Zee Maps, etc. You can also look on  Programmable Web for other geocoders. Depending on the size of your dataset, you could even go through the geocoding processing manually (i.e. typing in addresses and then noting their Latitude / Longitude), but is only useful for completely static data. Once geocoded, you can use a mapping API to fairly easily get your data to show up on a map.

To hand in on T-Square





Topics: Assignments | No Comments »

Reuters’ Open Calais

By Nick | February 13, 2008

Reuters release an new API called Calais based on a semantic entity extraction engine they acquired from Clear Forest last year. It can extract Entities, Organizations, Companies, Events, and relationships between these things. Read Write Web posted a good summary of the API’s capabilities here.

Thinking about how to integrate the capabilities of this API into news analysis or consumption interfaces would an excellent starting point for a final project.

Topics: Interesting | 3 Comments »

Info Viz Links From Class

By Nick | February 12, 2008

Here is a list of some of the Visualization examples shown in class today:

Topics: Interesting | No Comments »


« Previous Entries Next Entries »