Description:agRecommender can be used to estimate recommendations using rating data as input.
Abstract:agRecommender can be used to estimate recommendations using rating data as input. It is based on Apache Mahout and it provides several parameters for the recommendation algorithm. This module can be used by portal owners that would like to provide recommendations to their users but also by data scientists and researchers that would like to perform evaluations. In agINFRA the agRecommender is accessible by a RESTful interface.
In agINFRA project we developed a recommender command line interface that helps the developer or researcher to build a simple recommender and either get recommendations for the users of your dataset or perform simple MAE (Mean Absolute Error) evaluation. The first version that was ported on the agINFRA grid provides recommenders based on Collaborative Filtering and more specifically UserBased recommenders provided by Apache Mahout framework.