Class Projects - Fall 2014


Creates a streamlined way of viewing the news preferences of local audiences.

This is "LocalPulseDemo" by Northwestern U. Knight Lab on Vimeo, the home for high quality videos and the people who love them.
What it does

LocalPulse is about creating a streamlined way of viewing the news preferences of local audiences. Our current culture values the notion of sharing through various platforms of social media – this includes sharing photos, comments, opinions, literature and events. So, we wanted to analyze what was being shared in the context of news: what links to articles and media do people feel are important. We wanted to see whether there was a difference between what people were sharing in Chicago and what they were sharing in New York, what people are sharing in SF to Houston. Are there local trends? And finally, we wanted to bring that back to a national level and do a comparison between the national preferences and local preferences for further analysis.

Right now, on Twitter, there is a way to view trends geographically-- but these are general trends, and not necessarily news-related. Media consumers can utilize LocalPulse to view the “trends” in the scope of news articles and multimedia, which thus allows them to understand the “news culture,” geographically. Our user is an avid consumer of online news: someone who deeply cares about his own news intake as well as the journalistic preferences of his neighbors to better understand their interests. LocalPulse product allows users to see the disproportionality in content consumption by displaying the articles that are being shared via Twitter at a higher rate than the articles being shared across the nation. Furthermore, they can analyze the differences or trends with other cities to view other metropolitan snapshots of news preference.

How it works

Using MongoDB and the Twitter API, we are able to gather Tweets from our major metropolitan areas. By giving the Twitter Streaming API a set of coordinates, we are able to get access to tweets within a certain region. Using these Geo-tagged tweets, we count the number of times a certain news article is shared in a region within the past 24 hours. We also note how many times a certain news article has been shared nationally. This is simply the sum of the shares across all metropolitan areas. All of this information is stored in our Mongo database to be used by our website.

When a user wants to see the news stories from a specific city, we use the Flask web framework to access our database of links. Under the NationPulse column, we display the stories that are shared the most across the entire nation. For a specific city, we show the news articles that are shared disproportionately more in that city compared to all other cities. We also display this local share percentage. For example, in the above image, the news article was shared 67% of the time in Houston and 33% of the time in all other cities.

We then follow the links we chose to display to find relevant information about the news article. We are able to display the title, a preview of the article, and a picture associated with the article. When the user clicks on a link, they will be redirected to the source of the news article.

Key Technologies:

  • MongoDB
  • Flask Web Framework
  • Twitter Streaming API

Next Steps

When considering future work with LocalPulse, we would like to work on current shortcomings in the system and consider different users. A central problem with LocalPulse is the lack of links we are able to find using only Twitter. The main way to fix this is to simply consider more sources. If we considered Geo-Tagged posts on Facebook or Reddit the size of our database would certainly grow.

We also would like to shape LocalPulse to be useful for publishers. In the future, we intend for publishers to use this tool in two ways: they, too can view trends of their audiences and better tailor their news content to that “target audience” or local readership. Audience understanding is extremely important for journalists and publishers because it creates successful content. For example, we saw that in Chicago, many individuals care about sports and are sharing links to related content, which could indicate how publishing companies could potentially allocate resources. Additionally, it allows publishers within the same location to view their competition-- they can see whether their own content is trending, or someone else’s, and specifically, who that someone else is.

Figure 1.1: Screenshot of website with “Houston” tab selected

Figure 1.2: Screenshot of one of a display box, highlighted a trending article. The headline, a brief description of the article and a source are included. “The Local Share Percentage” corresponds to the proportionality of that link’s shares compared to the rest of the nation. 67% would indicate that the story is being shared 67% more in Houston than other cities.


Student Team:

Alex Kuntz

Priyanka Mody

Nicholas Scoliard

Faculty Guidance:

Larry Birnbaum

Rich Gordon