I need to process a database in order to add meta-information such as td-idf weights to the documents terms.

Successively I need to create document pairs with similarity measures such as td-idf cosine similarity, etc...

I'm planning to use Apache Lucene for this task. I'm actually not interested in the retrieval, or running a query, but in indexing the data and elaborate them in order to generate an output file with the above mentioned document pairs and similarity scores. The next step would be to pass these results to a Weka classifier.

Can I easily do it with Lucene ? thanks