wiki:WikiStart

Version 11 (modified by kiesel, 13 years ago) (diff)

search

About Skipforward

Skipforward is a distributed recommendation system using a lightweight ontology approach for formalizing opinions about item features. Items can be things such as songs or board games; example item features are the genre of a song or the degree of chance in a board game. Every user of the system is free to add new items and statements about existing items to the system. Naturally, opinions may differ between users---the system even encourages people to express dissent by supporting negation for item features. Skipforward allows discussions for any item feature as well as displaying these discussions in a way similar to web forums.

Skipforward uses the ​GNU Affero General Public License version 3.

Features

  • supports ontology-based subjective personalized annotations using the Skipinions ontology
  • implements personalized views using weighted aggregation of multiple statements concerning one item and feature type
    • competence metric works per feature type and is based on modified Pearson correlation
  • recommender functionality for...
    • similar items (based on their annotations)
    • items matching a predefined annotation profile
    • item annotations (recommending feature types highly relevant for the item recommender but not yet present for the current item)
    • experts (users with high correlation concerning one feature type)
  • explanations available for recommendations
  • personalized RSS newsfeeds
  • all information available as Linked Open Data, too
  • distributed design based on full replication using ​XMPP
  • integrated with DBTropes

Starting Points

  • ​Demo installation - "Login as demo user" logs you in read-only.
    • GettingStarted - how to log in and set up the system using an own XMPP account.
  • ​SystemWalkthrough - a screencast showing some of the major points of Skipforward, based on revision 224 (2008-10-26).
    • Recent features not included in the screencast: Trust network and opinion/user similarity metrics, alternative user interface, RSS feeds, improved search.

Development

Contact

Publications