lucene vs solr scoring

Question

Can some one explain (or quote a reference) to compare the scoring mechanism used by SOLR and LUCENE in simpler words. Is there any difference in them;

I am not that good at solr/lucene but my finding showed as if they are different.

P.S: i just tries a simple query like "+Contents:risk" and didn't use any filter other stuff.

Answer 1

Lucene uses concepts from the Vector space model to compute the score of documents. In summary, queries and documents can be seen as vectors. To compute the score of a document for a particular query, Lucene calculates how near each document's vector are from the query's vector. The more a document is near the query in VSM, the higher the score. You can have more details by looking at Lucene's Similarity class and Lucene's Scoring document.

Answer 2

The actual formula can be found in the Similarity javadocs.

Here's a summary of the parameters involved and a brief description of what they mean.

Solr uses Lucene under the hood, and by default Solr uses the default Lucene similarity algorithm.