Contributing

More RSS feeds are always useful. If there’s something specific you want to see, just add it in and open a pull request with the source’s raw XML RSS feed, a unique source ID, and a label indicating whether the source is “international” or “local.”

We face a tradeoff between seeking the broadest geographic coverage we can get (meaning including every local paper we can find) and accuracy and relevance (which would lead us to include only large, well-known, and high quality news outlets). We’re trying to balance the two objectives by including a third column indicating whether the source is one is a wire service, a dependable news source with solid international coverage, or a local source that may contribute extra noise to the data and may require specialized actor dictionaries. The distinction between the latter two is hazy and requires a judgement call. Eventually, these labels can be used to build event datasets that are either optimized for accuracy and stability (at the cost of sparseness), or micro-level, geographically dispersed (but noisy) coverage.