Recommendations

Engines

A recommendation Engine is identified by its unique Engine Name and is primarily characterized by its Engine
Type. The Engine Type defines which recommendation algorithms and which processing steps (e.g. filtering) are
used.

• Similarity: Similarity Engine recommends Assets which are semantically similar to a seed asset
  provided. Optionally the preference profiles of the user are taken into account too to provide
  personalized similarity recommendations.
• Preference: The Preference Engine recommends assets matching a preference profile, which is derived
  from user ratings. Multiple profiles per subscriber are supported
• Social Engine: recommends assets based on the subscriber’s social network profile (e.g. Facebook
  profile)
• Collaborative Engine: recommends assets based on collaborative filtering (Subscribers who watched
  this, also watched that…)
• Statistical Engine: recommends statistically relevant assets, e.g. most watched or best rated assets.
  Optionally the preference profiles of the user are taken into account too to provide personalized
  statistical recommendations.
• Moderated Engine: recommends assets which have been composed by an editor of the customer.
  Optionally the preference profiles of the user are taken into account too to provide personalized
  editorial recommendations.
• Search Engine: offers free text and structured search in the Asset Store. Optionally the preference
  profiles of the user are taken into account too to provide personalized search results.
• Posting Engine: recommends assets based on the subscribers postings to social networks (e.g. Tweets).
• Serendipity Engines: serendipity flavors of the preference and of the collaborative engines provide
  “surprising” recommendations, i.e. recommendations which to not perfectly match the user’s
  expeectations but which are expected to enrich the user experience.