This is a beautiful piece of work:
uBioRSS application catalogs RSS feeds. Feeds can be categorized within multiple classes, for example the JCR subject categories. Each feed will also be assigned an update schedule based on the source update frequency (hourly, daily, weekly, etc.) In the first step the nameCrawler follows each RSS link back to its source HTML/XML output.
2. findiT algorithms examine both feed and link text and identify all scientific names. Names are reconciled with NameBank identifiers and lexical and nomenclatural concepts to aid in integration with authority sources.
3. classificationSearch SOAP calls superimpose the names index with classification hierarchies to provide a navigational framework to the RSS articles.
4. LinkiT provides dynamic contextual links to authoritative data sources by maintaining name/key indicies to these sources.
5. RSS output is organized using taxonomic inteligence to provide contextual browsing by taxon. Output can be via RSS, email, or web page.
6. Users can create custom profiles to specify what sources they wish to review and can add feeds not currently provided. Regional or specialized taxa lists (endangered species, etc) can be uploaded for more specific monitoring.
7. LinkiT can dynamically embed hyperlinks to authoritativedate sources within the RSS, email, or HTML output, effectively 'advertising' the location of expert data related to taxa annotating the RSS output.
via Ingenta blog All My Eye - ALPSP Technology Update: Web 2.0 Hip or Hype?