Once your account is setup, there are only 4 steps to make any Media Library discoverable:
1) Select Attention Data
Both explicit (e.g. star rating) and implicit ratings (e.g. repeat views,
click-throughs, etc.) can be used. Discover My Network automatically trains
with each type.
2) Upload list of User Ratings
Upload the history of that attention data. You need to do this only once.
More on User Ratings...
Anatomy of a User Rating
A User Rating is simply made up of a
User ID
, a
Rating
, and an
Item ID
. For example:
<votes>
<vote>
<userID>5392003</userID>
<rating type="stars">5</rating>
<itemID>592003949212</itemID>
</vote>
...
</votes>
A User Rating can be any type of attention data -- from explicit "votes" or purchases to implicit actions like
clicking on details, watching a video to the end, etc. Multiple types of attention data can
be combined by the service to improve both recommendation quality and breadth of content.
Less...
3) Request Recommendations
Simply query the API with a UserID and any ANY arbitrary set of content (e.g. "Dance & DJ" genre).
The service will return the TOP recommendations for that user in milliseconds.
More on filtering...
Pre-filtering
We leave it to you to pre-filter the media you wish to recommend to meet your particular application.
For example, a user clicks the "comedy" videos link: you then give the service the list of comedy videos
and it instantly recommends the funiest videos for that particular user.
We have category provisioning to make it easy to request recommendations for large categories of items.
Less...
4) Submit New Ratings
To keep the recommendations fresh and constantly improving, just submit any new attention data as it happens.
The system re-optimizes the whole HumanNet with every action, improving the recommendations for every
user in the system progressively, independently and automatically.
