The mission of HMD is to provide a social web service for tracking estimates of the scope of political violence around the globe as it happens in the hope of informing policy and public discourse.
How many people have been killed, maimed, orphaned, displaced or otherwise grievously hurt by political violence in a given conflict, over a given time period? For instance, how many people were killed near the Raba'a al Adiwiya mosque in Egypt on August 14, 2013? How manu civilians were hurt in the violence following the 2009 Iranian elections? These numbers are difficult to come by. The definition of political violence is not widely agreed upon and there are no definitive official sources of data to estimate its scope.
Governmental organizations are not good at tracking conflict data because casualty estimates carry political costs that governments are not willing to bear. There are well known academic research efforts, such as the Uppsala Conflict Data Program (UCDP) at Uppsala University in Sweden, and the Center for International Development and Conflict Management (CIDCM) at the University of Maryland in the United States, that track casualty estimates, among other variables, on an annual basis for conflicts that meet their definitions of political violence. The UCDP data is searchable and available online for many conflicts through 2007 (as of the summer of 2010). The CIDCM data is published in biennial reports, the most recent of which (in 2010) can be purchased for $24.25 on Amazon and tracks data through 2007.
In addition, there are well-known non-academic and non-governmental organizations (NGOs), such as Amnesty International, that track conflicts and their casualties and provide data via reports and news stories.
All of this work, however, is contained in a diffuse body of web sites, agency and academic reports that struggle to keep pace with the evolution of political conflict across the globe. HMD's goal is to provide a way to organize data from these myriad sources, as well as from social media, to track the scope of casualties from political violence in as close to real-time as possible.
HMD is not a political blog. HMD does not advocate for any side in a particular political conflict. HMD does not filter content on this site with an editorial eye to slant coverage one way or the other. While HMD was created to focus attention on highly complex and controversial topics surrounding the casualty estimates that comprise the site, it is important for HMD to remain impartial. HMD is an information resource sustained by its users to help inform its users.
By aggregating publicly available data sources with reports culled through social network channels like Twitter and Facebook, HMD tries to provide the best casualty estimates in as close to real time as possible so our users can inform their personal opinions and public policies on global political conflicts.
Consider the June 2009 Iranian election and the violence that followed. There is scant information on casualty numbers, and the estimates that exist are divergent. The Iranian government has admitted to 36 deaths. CNN reported unofficial estimates of over 150 deaths in the protests on just one day (June 20th) of the protests.
Use example of 2008 war in Gaza and Iranian political violence to make the point that it's difficult to obtain estimates, but still important to do so. In the case of the Gaza War, the facts are disputed. The UN has its numbers. Various NGOs have their numbers. The PA has its numbers. The IDF has its numbers. And there is variation, even wide variation in these estimates. But all estimates point to well over 1000 Palestinians killed in the 3 week war, compared with 13 Israelis. That's a stark fact that should give someone, somewhere food for thought, whether they be a policy maker or a just someone talking to a friend over dinner.
Basic Model:
There's an OLTP model to capture reports of estimates. This model captures as much metadata as possible both about the source of the report as well as the casualty report itself. It captures these reports from the following channels:
web - a web form on hmd.org api - postings to a restful api email
- an email to [email protected] twitter - tweets @hmd.org facebook
- postings to hmd page and culled from member pages user feeds - rss feeds of users
On channels other than the web and api, we use microformats to embed report data in the channel. So we parse the email or tweet or fb posting or rss feed. If the posting is from a recognized account, we mark it as authenticated. Otherwise, it is considered unauthenticated. Unauthenticated reports can be viewed and summarized just like authenticated ones. But b/c we track it, they can be filtered. Reports that are initially unauthenticated can later be authenticated by a user adding their channel to an authenticated account.
Another kind of transaction we process online is voting for reports. This allows users of the site to reveal the degree of their interest in particular reports. It allows HMD to rank reports and show trending issues.
Another kind of transaction we process online is commenting on reports. Comments are limited to 1024 characters, which is enough for about 14 average english sentences, or 1-2 paragraphs. The purpose of the limit is to prevent the site from becoming a wiki about specific political conflicts and try to focus people on the reports and the immediate issues surrounding them.*** WHAT DOES THIS MEAN?
We also have OLAP model. This model is a basic star schema with the main fact table being the casualty table. It has dimensions around it involving metadata about both the casualty estimate itself (when and where it occurred, what kind of casualty report (death, injury, displacements, etc), what conflict it is associated with, which side of the conflict the casualty is associated with, etc) as well as the reporter of the estimate (channel, when/where/who report came from, is it authenticated, number of votes, number of comments).
The OLAP table is what drives the output side of the web site. People can define OLAP cubes that can be accessed by any Mondrian front-end, like JPivot or OpenI. These cubes further, can be shared, rated, commented on, etc. We provide several cubes ourselves that are always available to people. Analysis from popular cubes are shown prominently on the HMD site.
casualty table is main fact table dimensions include: source of report time of casualty location of casualty conflict
create table casualty ( estimate int unsigned source (who reported it) time (when did it happen) location conflict_id
)
dimensions time location