content based recommender in hadoop

Question

I am thinking about implementing a content based recommendation project idea in hadoop.

Just wanted suggestions if weka can be used with hadoop or if it is advisable to use Mahout for a cbf based implementation

Please advise

Thanks

You could get much more valuable feedback, if you could be a little bit more specific on what exactly you need (which algorithm, what kind of data, etc.) — vefthym, Aug 19 '14 at 12:37

score 0 · Answer 1 · answered Aug 19 '14 at 14:58

What is your content? What data surrounds or makes up your items?

There are several ways to use this for recs and some can blend in collaborative filtering data if appropriate. Take a look at rowsimilarity in Mahout. It calculates the similarity of an item to all others. If you input content in a way it can digest you'll have a CBR.

For instance input for recommending similar blogs posts would be: postID, tokens-from-the-text-of-the-post

Then for each postID you'll get a list other postIDs that use similar words. You can do the same for tags, categories, or other metadata too.

content based recommender in hadoop

1 Answers1