The Tagger project is a library making it possible to extract relevant tags (keywords and named entities) from texts.
Named Entities are rated for relevancy according to their position, frequency and whether they are enclosed in specific HTML-tags (<emph>, <strong> etc.).
The library can be included in a webservice wrapper (https://github.com/40c/tagger-webservice) or it can be integrated into your favorite CMS (e.g. Drupal: http://drupal.org/project/tagger)