Mendoza

Mendoza is a tool/framework for generating and presenting recommendations. End users enter information themselves and the tool will return a ranked list of items that the user is likely to appreciate. Items such as movies, songs, directors and artists are supported out of the box. Furthermore, the application is extensible so persons with a working knowledge of the Java programming language can devise their own engines and datatypes. Download ZIP or run via Java Web Start.

Create new document and select recommendation type(s).
Create new document

Thesaurus recommendations based on strings (i.e. words). Add one via context menu of 'source' table.
Add source word

Word is added to table.
Added word: 'good'

Assign a weight in order to generate recommendations.
Assigned weight to 'good'

Mendoza supports multiple receommendation types. Below example uses lastfm API.
Create new lastfm document

Import a specific user's 'loved' tracks.
Import 'loved' tracks

Imported 'loved' tracks

Assigned weights to songs