Mahout Recommender - questions to setup user preference -

i'm looking advice / guidance --

i'm working on recommendation engine / personnel assistance app, using mahout framework -

what want new users of app begin answering 5 questions , use answers questions effect recommendation -- pretty feeding answers user-preference

i'm not sure how incorporate code, i'm not sure begin looking - i've been googling none of search results address this...

any suggestions / advice / guidance appreciated

thanks

i did new spark itemsimilarity implementation year ago. you'll need search engine recommendations query because mahout doesn't have server. i'd suggest using new "universal recommender" engine template predicitonio. uses mahout calculate model , elasticsearch serve it. https://templates.prediction.io/predictionio/template-scala-parallel-universal-recommendation

preditionio framework of integrated components provide event server (for event storage) integration hadoop/hdfs, spark, hbase, , rest or sdk api. install , template plugin engine. provide pretty advanced recommendations queries multiple event ingestion, hybrid content-based method tune results, , several methods of using popular items backfill when no other recommendations can made. uses realtime user actions recommendations.

this last bit super important if want have users go through training. way see benefit of training in realtime. check site, did talking about: https://guide.finderbots.com notice "trainer". presents movies , asks thumbs or down many care do, when ask recommendations based on realtime preferences of user. need create account first have user-id.

the way created list trainer cluster popular items. clustering mean based on users preferred items. clustering produces items differentiated because belong different clusters, means different user-sets tended them, , popular ones more known users when go through training. these things have in trainer.