Calculating cosine similarity in mahout

Question

In order to find the similarity between two documents , i am planning to adopt the use of mahout to perform this task .

The process would include :

1. converting the doc to tf-idf
2. Removing stop words (making the search effective)
3. Run cosine similarity
4. Give the degree of similarity

I am planning to implement this in mahout . I am a beginner to mahout , can somebody help me out with a few tutorials to perform this and tell me if this is a effective means to calculate the similarity between the documents

Answer 1

You don't need to implement anything. Use seqdirectory and seq2sparse to vectorize your data. After that you can use RowSimilarityJob to compute pairwise cosine similarities.