MEAD add-ons

In each section, you should download the corresponding *tar.gz file. If you prefer, you can instead download some individual scripts extracted from the *tar.gz file

Preprocessors for converting text and HTML files into docsents and clusters for use with MEAD
CIDR for clustering documents
Evaluation (combined)
- IR-based
  - for calculating kappa from 3 or 4 extracts, you'll need this and that.
  - Pearson product moment
  - Spearman's rank correlation coefficient
- Content-based script and modules
Rerankers (combined)
Utilities for sentrels
DUC scripts (combined)
- Creating sentjudges from manual summaries 2002
- Creating sentjudges from manual summaries 2003
- Creating docsents from DUC data, and you'll also need this.

For security reasons, we've changed the names of the individual scripts from .pl to .perl. If you download the .tar.gz files, the .pl files remain properly named.

Credits:

Tim Allison: current scripts for conversion of HTML and text to docsents and clusters
Ali Hakim: made CIDR more user-friendly
Anna Osepayshvili: MMR rerankers
Hong Qi: all DUC scripts
Dragomir Radev: CIDR, pearson and spearman scripts
Horacio Saggion: content-based evaluation script and modules
Simone Teufel: scripts for calculating kappa over multiple extracts
Adam Winkel: original scripts for conversion of HTML and text to docsents and clusters
Zhu Zhang: part of the content-based evaluation script and all of the time-based, source-based and cst-based rerankers, as well as all sentrel utilities