"Fractor is a social venture and media company dedicated to encouraging civic participation through open source online and mobile technologies. Fractor's initial development is a simple, groundbreaking web application that matches news stories with opportunities for service." (more at http://www.fractor.org).
In the fall of 2008, I developed a cosine-similarity content recommendation system that matches news stories to relevant actions. This involved writing a probabilistic parser to scrape news article text, as well as integrating natural language processing tools and a document vector space model to analyze text and extract semantic content. At the end of spring 2009, I initiated the development of a system based on Latent Semantic Analysis for more sophisticated recommendations.
Cropped screenshot of news-action recommendations from Fractor demo site. (Image courtesy of Fractor Corp. All rights reserved.)