Object of class ServerRecommender

Share:

Description

The core implementation of Recommender and furthermore MyrrixRecommender that lies inside the Serving Layer.
This is the recommendation engine class.

The ServerRecommender has a local model, allowing it to build recommendation models based on data which are locally stored on your disk.
Next to the local serving, it also allows to build recommendation models based on data which is distributed on Hadoop. Special out-of-the-box classes exists to let it run on CDH, AWS and on Hadoop clusters. If you run the ServerRecommender in a distributed mode, we assume that you have set up the Computation layer already. This R package allows to ingest new data, update the model, get recommendations and similarities based on the recommendation engine which is running.

Arguments

bucket

character string with the bucket that Serving Layer is using for instances

instanceID

character string with the instance ID that the Serving Layer is serving. May be 0 for local mode.

localInputDir

character string with the local input and model file directory

partition

integer with the partition number in a partitioned distributed mode. 0 if not partitioned.

allPartitions

reference to an object that can describe all partitions; only used to get their count (seee http://myrrix.com/docs/serving/javadoc/index.html)

Slots

recommender:

A java object of class net.myrrix.online.ServerRecommender

Examples

1
2
recommendationengine <- new("ServerRecommender", localInputDir=tempdir())
recommendationengine