7.2.2 The BM25 Ranking Algorithm
BM25 extends the scoring function for the binary independence model to include
document and query term weights. The extension is based on probabilistic
arguments and experimental validation, but it is not a formal model.
BM25 has performed very well in TREC retrieval experiments and has influenced
the ranking algorithms of commercial search engines, including web search
engines. There are some variations of the scoring function for BM25, but the most
common form is: