Category Archives: research interests

Unfluence Project Launched!

UnfluenceLogo
Greg Michalec and I just launched a website that allows users to interactively explore political contribution data as a network map. We won first prize in the Sunlight Foundation Mashup competition! Try your own query at http://unfluence.net/

UnfluenceExampleSmall
This was done as an entry in the Sunlight Foundations Mashup Contest. Cross your fingers that we win and get some resources to continue the project! ;-)
Continue reading Unfluence Project Launched!

API access to US state political funding data

MoneyInStatePoliticsLogoThe National Institute on Money in State Politics now provides a service where users can submit queries and download data in xml format. The documentation for the service (which requires setting up an account login) is here.

They also include services for generating various graphic reports including a Political Contribution Logorithmic Scatterplot Profile with interactive javascript inspection of candidates.

StatePoliticsPULSEChart

I did quick cut-and-paste from another site which uses thier data, and was able to generate network maps of the top tend funders for a few california candidates. But with access to the full DB via the API, it is possible to do much more sophisticated maps and even animations – for most states they have data going back several years.

CaliContributions

Political Network Datamining

Its not what they say during the campaign that counts…

..its what they do in office. The nice thing about politicians is, a lot of what they do is a matter of public record, increasingly available on the internet for free. Wouldn’t it be great if ordinary people had the tools to make this political landscape visible?

I’m posting here some examples I’ve come across of intriguing maps and representations of political data. Mostly these fall under the broad heading of “mapping congress” based on data available from the THOMAS db of congressional record.

Committee Interlocks

I found a nice preprint in arXiv that explores the hierarchical relationships formed by the pattern of overlap of Representatives in committees:

commStructFig2Larger
Continue reading Political Network Datamining

A Call for Collaboration (0.1)

I’m looking for people to team up with to build a non-profit commons based R&D organization. Here’s the current state of my evolving pitch:

Commons Business Model (long term)

CommonsBassedResearchLogoI see this as a project running on a couple of different time scales with various goals. The broader and longer term of which is to test and demonstrate the feasibility of an R&D business model based around intellectual commons. So organizations, academic institutions, and companies would pay for expertise and specific tool delivery, but intellectual property is drawn from and returned to the public domain. Clients would be buying the innovation, the implementation (that final 20% it takes to hook up all those great free libraries into something useful) and saving big bucks by not re-writing everything from scratch. This kind of model is especially appropriate for academics and NGOs which are (nominally at lest) working for a common good rather than directly competing for monopoly.
Continue reading A Call for Collaboration (0.1)

Extraction, Visualization & Analysis of Corporate Inter-Relationships

Interesting project EVA with goal of creating a semi-automated system for parsing and extracting corporate relationships from SEC 10-K filings, ‘tho they did use some existing ownership dbs as well. Software uses a nice technique of identifying paragraphs likely to contain owership information and presenting to user for verification. Seems that much of the work was done in 2001, I can’t tell if it is an ongoing project.
Continue reading Extraction, Visualization & Analysis of Corporate Inter-Relationships

Mapping Congress

It is still very rough, but I’ve been having a very interesting time looking at contributions among PACs in FEC Data. This image shows transactions to Senate candidate’s central committes that were reported in May of 2006. Colors by party, grey is unkown / unspecified. Labels are on candidate’s committes, but some are violently truncated. Names of PACs have been removed to protect the guilty (actually because they are too long and make the graph cluttered)
senatePacsMay2006

They layouts are not converging very well, so the structure is not that accurate. This is also only showing one month’s data, so not very representative. But need much more specific queries and good aggregation rules to deal with longer time periods. Easy to imagine making it interactive, zoomable, animating the time data, adding to SNA metrics, etc. A long ways to go to build software to do this, but I’m excited to have something to show!

Newbie’s view of a Network

I started poking around in omidyar.net a few days ago, curious if it can be a resource for locating activist techs and funding. As I’m not really a member of any of the “social networking” sites, I don’t really have a sense of the social protocol. It is kind of like stepping blind into a cocktail party. A few people immediatly contacted me, but I didn’t know who they were other than by reading their profiles. So I got curious about the structure of communication of the site. After a day or so of coding with a little help from carnivore, htmlparser and sonia, I’ve got a few pictures of the network of positive feedback on comments. (people have the ability to give points to eachother’s posts in the fourms, kinda like slashdot)
snapshot of newbie's view of omidyar feedback network

Continue reading Newbie’s view of a Network

Commerce info tracking, etc

Nice idea, building collaborative software for tagging and tracking commercial products. “Network of Integrated Consumer Knowledge” Eugene based software co. http://www.grasscommons.org/

Grass Commons, a 501(c)3 public interest charity, is building pipelines between those who can generate information about products and companies and those who can use that information to build a more sustainable economy and a better world.

Grasscommons seems to have a collaborative relationship with Hooze,

Hooze.org is about collaborating to gather useful, reliable info about the products and companies that are shaping our world.

Hooze is the world’s first public wagging site. Wagging combines wiki and tagging in unique ways so that communities of users can organically develop ways of organizing and presenting information. The combo cards at the bottom of this page are an example of how wagging differs from other tools.

And also seems to be building tools like data integration engines. Not only that, but also researching for open-source licensing for structured data. Something that would be very important for an organization that would like to curate data and provide it to the public…

CSDE: Techniques for Visualizing Transmission in Dynamic Networks

simMovieImgDr. Martina Morris (PI) & Dr. Mark Handcock at the Center for Studies in Demography and Ecology of University of Washington are funding a SoNIA-related project contract to integrate dynamic network visualization techniques with the R statnet package developed by Morris, Handcock, et al at the CSDE. The funds come from the NIH grants supporting the Network Modeling Project at the University of Washington (grants R01 HD41877 and R01 DA12831)
The focus is on adapting and developing visualization techniques for dynamic network data. The specific emphasis will be on techniques relevant for understanding:

a) infectious disease transmission (change in state of elements due to a network diffusion process) and

b) the stationary dynamics (model-based addition and removal of nodes/edges) of longitudinal network data and simulation output.

The goal is to develop robust, rigorous and repeatable procedures for visually interpreting time-based network data. We will achieve this by linking existing software components and improving existing techniques to generate animations and export movies from R or other statistical packages into standard formats suitable for use in presentations or websites.