Compute the BM25 score of a new document

[wuchen95]

Hi, I wonder if I can compute the BM25 score between a query and a new document which is not in the collection index. I wanna use the statistical information from the collection to compute the BM25 score of this new document.
However, the current interface takes into a 'docid' which means the document is already in the collection index.
Maybe I could insert the document into the index and then do the above operation. However, I do not want to index all the collection (e.g., MS-MARCO doc is large) again for just adding a single document. So is there an incremental index interface like ES?
Or how can I better solve the problem? Thanks!

[lintool]

Hi @wuchen95 - currently, the feature of incremental indexing from Lucene is not exposed in Pyserini, so to compute the BM25 score of a new document, you'll have to compute it yourself from the component tf/df and other stats.