shows you a visualization of how your candidate/representive (the center circle) is represented in the media.

By default, the system searches and analyzes all the news sources available and returns the results found. However, you may select the dates and sources you wish to filter by clicking and

By hovering over a circle, the number of co-occurrences between that entity and the central candidate/representative will appear.

For example, in the above left image, we find Rick Perry and Garnet Coleman were mentioned or "co-occurred" in the same sentence, or within 3 sentences, 139 times overall in the articles found for Garnet Coleman.

By clicking on the center circle in the visualization, you'll see information related to your candidate and additionally, a table of the results, which includes clickable links and filtering mechanisms. For example, in the above right image you can filter by entity type by clicking on any of the headers: POLITICIAN, PERSON, ORGANIZATION, MISC, BILL.

The number of nodes presented in the visualization is controlled by clicking on the following to show more or less nodes, or full details which wil display all additional nodes.
At any point you may return to the initial results/controls screen by clicking on the following:

By clicking on any other circle (also called an entity), the articles between that entity and the central candidate appear.

For instance, to the right, we see the information that appears aftering clicking on the node for Jessica Farrar.

We see that she occurred in 37 instances over 27 articles with Garnet Coleman, and fifteen of these articles are from the Houston Chronicle (HC), one is from the Dallas Morning News (DMN), etc.

After a short description of her, the article urls and text snippets along with highlights are presented. Again articles may be filtered by clicking on the news source desired.
News: See our project proposal page along with several other great ones for the Knight Foundation News Challenge

Voter turnout at the state level is very low for offices which hold great power over a citizen’s daily life (the current governor of Texas won by a landslide and was elected by less than 20% of registered voters) and a large part of that is due to voters not feeling that they have sufficient impartial knowledge regarding the candidates. Voters may find it difficult to trust ads placed by politicians, regardless of party affiliation, are unlikely to read or trust news coming from sites they don’t already visit, and even if one has the time to keep up with the news, it is impossible to find and read all the articles referring to the candidates.

Why is our approach innovative?

In particular, we would like to know more about the connections of a politician with other colleagues, businessmen, companies, organizations and other relevant third parties. By doing this, we could learn more about their political agenda and their lobbying activities, and quickly understand who they serve, who they collaborate with and who they oppose.  In summary, we are producing a social network that captures the relationships between the different actors of the political landscape in a way that leverages any publically accessible data someone wishes to incorporate.

How can we come up with these social networks?

Social networks have been used in a variety of settings for understanding complex social systems. The problem is coming up with methods to automatically generate these networks from the vast amount of data we are exposed to everyday. is an automated web app tool to do the heavy lifting and provide you with a quick way to get a better understanding of who the candidates are for an upcoming election. In order to do this, we start with a seed list of candidates/representatives, and then search for them in major news outlets pertaining to your state (a preset list of newspaper sites, blogs, social media, and open gov sources running the political spectrum and including campaign contribution data and representative voter histories, or for more tech savvy users, a subset or list of your choosing).  We then create an interactive network visualization that will allow you to see how a given candidate is represented in the media by showing the discovered connections he/she has with other politicians, companies and organizations. Those connections that have more presence in the media, will be shown more prominently so that users can quickly find them. Furthermore, connections will be able to be expanded to show all the source urls and text snippets pertaining to it. This way, users will always be able to go to the media sources that lead to the relations in our visualizations.

What else would our tool provide?

Additionally, we will provide the capability to comment/discuss/enrich parts of our visualizations, thus involving the users in a fruitful way. We expect to engage our users this way so that our tool will not only be used as an information/educational site, but also as a forum for discussing aspects of our politicians’ associations. In a sense, we will provide a crowdsourced platform for the general public to include contrasting data about those relationships.

Finally, we plan to further analyze our generated networks. Leveraging data mining, information retrieval, natural language processing and big data network analysis techniques , we plan to discover distinguishing topics most associated with a given politician, and to find clusters/communities of politicians and companies that usually relate to each other.

What is our end goal?

We believe that a democracy is as healthy as the commitment and participation of its citizens in the political process. That is why we want our users to be more than consumers of information and to actively engage the political debate by providing the means for a focused discussion of the activities of their candidates

is a joint research project between the DAMA-UPC and LARCA groups being implemented at the UPC in Barcelona, Spain.  Team members include: