CODEWITHSUNDEEP
javamahout
user1067549
user1067549

Reputation: 41

mahout Recommender

The Apache Mahout Recommender Documentation mentions the following: 

// Construct the list of pre-computed correlations
Collection <GenericItemSimilarity.ItemItemSimilarity>  correlations = ...;

I'm not sure how the actual construction is done in the above line. Can someone provide an example?

ItemSimilarity itemSimilarity  =  new GenericItemSimilarity(correlations);

Upvotes: 3

Views: 1807

Answers (2)

Daniel Zohar
Daniel Zohar

Reputation: 2092

This example refers to the case where you have the similarities already computed, by the Hadoop job for example, and stored on the filesystem or database . As the constructor documentation reads:

A "generic" {@link ItemSimilarity} which takes a static list of precomputed item similarities and bases its responses on that alone. The values may have been precomputed offline by another process, stored in a file, and then read and fed into an instance of this class.

If you have tens of millions of recommendations or less, you can simply compute similarities on the fly and use the other GenericItemSimilarity constructor - GenericItemSimilarity(ItemSimilarity otherSimilarity, DataModel dataModel)

For example:

DataModel dataModel = new FileDataModel(new File("path://to/file.csv"));
ItemSimilarity itemSimilarity = new LogLikelihoodSimilarity(dataModel);
ItemSimilarity itemSimilarity = new GenericItemSimilarity(itemSimilarity, dataModel);

Upvotes: 3

Sean Owen
Sean Owen

Reputation: 66886

There's no magic here, it's just suggesting you create a bunch of those ItemItemSimilarity objects, one for each item-item similarity that you know about.

Collection<GenericItemSimilarity.ItemItemSimilarity> correlations = 
  new ArrayList<GenericItemSimilarity.ItemItemSimilarity>();
correlations.add(new GenericItemSimilarity.ItemItemSimilarity(1, 2, 0.5));
...

You can make it this way or any other way you want.

Upvotes: 2

Related Questions