• Sentiment-classified movie reviews
• Congressional floor-debate transcripts, with support/oppose labels
• AP88 data for some similarity-based pseudoword disambiguation experiments
• Multi-parallel proof/verbalization data for a project on verbalizing NuPrl mathematical proofs using multiple-sequence alignment
• Document sets used for ordering and summarization experiments

And here are some results from experiments.

• Extracted paraphrases together with human evaluation judgments, from a project using multiple-sequence alignment to learn paraphrases from comparable corpora.