如何使用Mahout计算用户和按标签分类的项目之间的对数似然相似度？

Question

I have read the Mahout in action but i'm not clear about Log-likelihood similarity. I have User and Item, classified by tag tag. And now i don't known how to calculate Log-likelihood similarity between User and Item. Anyone can help me? or give me some examples or demo about this?

Answer 1

You don't; you calculate similarities between users, or between items. Users and items aren't "similar" in this sense; you use a recommender to infer ratings or a strength of association, but it is not a log-likelihood similarity.

You use LogLikelihoodSimilarity to compute this metric, to answer your question.