Comment by garply

The bulk of it was:

News article title extraction. News article relevant thumbnail extraction. News article text body extraction. Generating publicly traded stock symbols from business news articles. Some Techmeme-style document clustering.