mchinnasamy / sparkreccengine Goto Github PK

Requirements:

1) Install JDK and set JAVA_HOME

2) Install Spark 1.6.2 and set SPARK_HOME

3) Include SPARK_HOME/bin in your PATH variable

4) Install sbt

Setup:

1) Clone this git project

2) You may use the provided sample ratings file in the data folder. Simply unzip both the files and use mongoimport to load data in to mongodb

Getting recommendations:

1) If you follow the instructions from setup above, you will have review_ratings details in a collection. 
   Otherwise you will have to provide your ratings in a 'review_ratings' collection

2) If required, you may generate a few sample user preferences (say 20) by creating random values for userId, businessId, review_stars fields. 
   Load this data in to a 'personal_ratings' collection

3) Submit recommendations program by executing submit-scala.sh script

   Usage: 
      submit-scala.sh [-h <host>] [-p <port>] [-d <db_name>] [-u <userId>]

4) Spark ML library uses ALS algorithm to generate recommendations and saves the output in a 'user_recommendations' collection

Questions or comments, you may write to [email protected]