Kruganmantimes.com Using Keyword Extraction

We recently discovered Kruganmantimes.com, a webpage that takes the recent version of the New York Times and replaces everything with Paul Krugman media. This project by software developer Vincent Woo uses AlchemyAPI’s Keyword Extraction feature to make the magic happen. Here is a short Q&A with Vincent Woo:

Where did you get the idea for The Krugman Times?

 
Foreign Policy asked me the same thing earlier, so I have an answer prepared:

"I sent a friend of mine a NYT article (this one) and told him I thought it was great article. He replied, cheekily, 'Great? But it's not by Paul Krugman!' From there it was a quick jump to 'What if Krugman wrote every article on the NYT?' and buying krugmantimes.com."
 

How long have you been a Krugman fan and has he seen this site?

I suppose I've been a fan ever since I saw Krugman decimate some British MPs in a Austerity debate, so about a year now.

Krugman definitely saw my site, and he even blogged about it! It was very satisfying, and responsible for pretty much all the traffic I got.
  

How does The Krugman Times work?

The Krugman Times is exactly one Node.js dyno running for free on Heroku, and stood up admirably to hundreds of concurrent users. It makes use of a free Redis addon for heavy caching of both the nytimes.com homepage and AlchemyAPI’s endpoints. Essentially, I pull down the nytimes.com page and do some DOM analysis to pull out stories. I send each story separately to AlchemyAPI for Keyword Extraction, and replace those with appropriately styled and cased economics gibberish. I also take the time to swap out images for pictures of that handsome devil, Paul Krugman. 

What are you working on now?

My day job is at everlane.com, where we are an online-only fashion retailer that uses the cost savings associated with not maintaining physical retail presences to put high quality clothing into the hands of consumers very affordably. I am also working on a Twitter bot for spamming people, but I don't think Twitter's gonna like that one as much.
 

Precog Uses Sentiment Analysis API for Oscar Winner Predictions

As the 85th Academy Awards draw closer, everyone (including AlchemyAPI staff) is placing their predictions. Precog is making it easier by analyzing Twitter mentions that mention #Oscars. They've used their Precog analysis platform to run the predictions and the ReportGrid API to create interactive visualizations.

Our Sentiment Analysis API was used in a ReportGrid Bar Chart showing positive Twitter mentions of the Best Picture nominees, shown below.

Precog also used our Sentiment Analysis API to answer the question: When are people the most positive and negative in their #Oscars conversations. You could use this ReportGrid Line Graph to figure out the best time to get your #Oscars predictions for the losers (negative sentiment) or the winners (positive sentiment).

For more information about Precog and their Oscar Winner Predictions, visit read their blog post: http://blog.precog.com/?p=403

Interview with Michael Lissack, the Executive Director of the Institute for the Study of Coherence and Emergence and ISCE Professor of Meaning in Organizations.

Once upon a time I was an investment banker and financed state and local governments. Upon my retirement (very premature at age 36) I became an academic and an entrepreneur.We founded ISCE in 1999 to study social complexity theory and its applications.

• What is the idea behind the ISCE library project?  

To create the library of the future. As specialized librarians become harder to find, this project aims to mimic a well-read expert in a particular topic. ISCE focuses on the Philosophy of Science and related topics, and can suggest pertinent materials that the reader should investigate as well as users with similar interests to communicate with.

• What problem is the ISCE library project solving?

As we become more reliant on the Internet we have found fewer and fewer opportunities to employ the well-read reference librarian.  There simply are too few people one can turn to with a snippet of an ill-formed idea and say "help me find more like this to read."  The ISCE Library and its sister projects (http://epi-search.com a generic research tool and http://systemspedia.org an encyclopedia of systems science and cybernetics ) are trying to become the Internet equivalents of that great reference librarian -- but one who is available 24/7 and supplies links in the flash of an eye.  A student, for example, can take an assignment, feed it to epi-search.com, discover great things to read, do their reading, write an outline, submit that to epi-search.com, find more things to read, read, write a first draft, submit that to epi-search.com and find things to read which will help make their project truly their own.

• How does AlchemyAPI help in your project?

We use the AlchemyAPI Concept Tagging functionality to analyze the text of each book in our collection. The Concept Tagging call provides the high-level topic information  we need to formulate the connections between books. Since two authors can use similar language but be describing completely different subjects, the Concept Tagging call is a more sophisticated way of understanding the topics in the text.  By combining the concepts call with identified keywords we can better zoom in on material which answers the question "find more like this."

• What are your future plans for the project?

We want to try other AlchemyAPI features, like Relations Extraction, Named Entity Extraction and some of the interesting features we've seen on the AlchemyAPI release schedule. We also want to release additional topic collections.  Right now, we are trying to gain traction with the ICSE library, and prove our model can be applied in a very general way. We hope people will try the general research tool at http://epi-search.com and see how it can help in their writing.

Anyone who has used Twitter in an effort to gauge the overall sentiment towards a particular candidate knows that the sheer number of tweets makes it impossible to do by hand.  Luckily Precog and AlchemyAPI solved that problem for you. We partnered up to launch a Real-Time Twitter Sentiment Analysis. Precog gathered tweets in real-time, then used our Sentiment Analysis API to score the tweets for Romney and Obama on a +1 (positive) to -1 (negative) scale, with the resulting analysis showing a state-by-state breakdown of Twitter sentiment.

Ever wonder how people in different states tweet about a candidate? Is it surprising that liberal Texans are more vocal on Twitter yet with Californians it’s the reverse? Ever wonder how the presidential conversations on Twitter changes from day to day? Answer your questions and generate new ones by rolling over each state here and viewing real-time sentiment scores.

In order to make this possible, first Precog searched for tweets using Twitter’s API and then analyzed these tweets using AlchemyAPI’s Sentiment Analysis API. This data is easily stored using Precog’s API and then Precog’s interactive development environment, Labcoat, is used to analyze the data. The resulting query was exported into a data visualization app that uses the ReportGrid visualization library. You can visitwww.precog.com/elections2012 to learn more and grab the code to embed the Geo Chart on your website.